An in-pixel histogramming time-to-digital converter(hTDC)based on octonary search and 4-tap phase detection is presented,aiming to improve frame rate while ensuring high precicion.The proposed hTDC is a 12-bit two-ste...An in-pixel histogramming time-to-digital converter(hTDC)based on octonary search and 4-tap phase detection is presented,aiming to improve frame rate while ensuring high precicion.The proposed hTDC is a 12-bit two-step converter consisting of a 6-bit coarse quantization and a 6-bit fine quantization,which supports a time resolution of 120 ps and multiphoton counting up to 2 GHz without a GHz reference frequency.The proposed hTDC is designed in 0.11μm CMOS process with an area consumption of 6900μm^(2).The data from a behavioral-level model is imported into the designed hTDC circuit for simulation verification.The post-simulation results show that the proposed hTDC achieves 0.8%depth precision in 9 m range for short-range system design specifications and 0.2%depth precision in 48 m range for long-range system design specifications.Under 30×10^(3) lux background light conditions,the proposed hTDC can be used for SPAD-based flash LiDAR sensor to achieve a frame rate to 40 fps with 200 ps resolution in 9 m range.展开更多
Ground penetrating radar(GPR),as a fast,efficient,and non-destructive detection device,holds great potential for the detection of shallow subsurface environments,such as urban road subsurface monitoring.However,the in...Ground penetrating radar(GPR),as a fast,efficient,and non-destructive detection device,holds great potential for the detection of shallow subsurface environments,such as urban road subsurface monitoring.However,the interpretation of GPR echo images often relies on manual recognition by experienced engineers.In order to address the automatic interpretation of cavity targets in GPR echo images,a recognition-algorithm based on Gaussian mixed model-hidden Markov model(GMM-HMM)is proposed,which can recognize three dimensional(3D)underground voids automatically.First,energy detection on the echo images is performed,whereby the data is preprocessed and pre-filtered.Then,edge histogram descriptor(EHD),histogram of oriented gradient(HOG),and Log-Gabor filters are used to extract features from the images.The traditional method can only be applied to 2D images and pre-processing is required for C-scan images.Finally,the aggregated features are fed into the GMM-HMM for classification and compared with two other methods,long short-term memory(LSTM)and gate recurrent unit(GRU).By testing on a simulated dataset,an accuracy rate of 90%is obtained,demonstrating the effectiveness and efficiency of our proposed method.展开更多
Skin cancer,a severe health threat,can spread rapidly if undetected.Therefore,early detection can lead to an advanced and efficient diagnosis,thus reducing mortality.Unsupervised classification techniques analyse exte...Skin cancer,a severe health threat,can spread rapidly if undetected.Therefore,early detection can lead to an advanced and efficient diagnosis,thus reducing mortality.Unsupervised classification techniques analyse extensive skin image datasets,identifying patterns and anomalies without prior labelling,facilitating early detection and effective diagnosis and potentially saving lives.In this study,the authors aim to explore the potential of unsupervised learning methods in classifying different types of skin lesions in dermatoscopic images.The authors aim to bridge the gap in dermatological research by introducing innovative techniques that enhance image quality and improve feature extraction.To achieve this,enhanced super-resolution generative adversarial networks(ESRGAN)was fine-tuned to strengthen the resolution of skin lesion images,making critical features more visible.The authors extracted histogram features to capture essential colour characteristics and used the Davies-Bouldin index and silhouette score to determine optimal clusters.Fine-tuned k-means clustering with Euclidean distance in the histogram feature space achieved 87.77% and 90.5% test accuracies on the ISIC2019 and HAM10000 datasets,respectively.The unsupervised approach effectively categorises skin lesions,indicating that unsupervised learning can significantly advance dermatology by enabling early detection and classification without extensive manual annotation.展开更多
Esophageal cancer is one of the most difficult cancers to treat since it is often at an advanced stage at the time of symptom presentation.For locally advanced esophageal cancer,treatment options include multidiscipli...Esophageal cancer is one of the most difficult cancers to treat since it is often at an advanced stage at the time of symptom presentation.For locally advanced esophageal cancer,treatment options include multidisciplinary treatment such as surgery or definitive chemoradiotherapy.Surgery has a high local control rate because it involves excision of the cancer along with the surrounding organs;however,it is still highly invasive,although advances in surgery have reduced the burden on patients.On the other hand,chemoradiotherapy may also be applicable in cases in which surgery is inoperable owing to complications or distant lymph node metastasis.However,chemoradiotherapy using X-ray irradiation can cause late toxicities,including those to the heart.Proton beam therapy is widely used to treat esophageal cancer because of its characteristics,and some comparisons between proton beam therapy and X-ray therapy or surgery have recently been reported.This review discusses the role of proton beam therapy in esophageal cancer in comparison to X-ray therapy and surgery.展开更多
Background:Pneumonia remains a critical global health challenge,manifesting as a severe respiratory infection caused by viruses,bacteria,and fungi.Early detection is paramount for effective treatment,potentially reduc...Background:Pneumonia remains a critical global health challenge,manifesting as a severe respiratory infection caused by viruses,bacteria,and fungi.Early detection is paramount for effective treatment,potentially reducing mortality rates and optimizing healthcare resource allocation.Despite the importance of chest X-ray diagnosis,image analysis presents significant challenges,particularly in regions with limited medical expertise.This study addresses these challenges by proposing a computer-aided diagnosis system leveraging targeted image preprocessing and optimized deep learning techniques.Methods:We systematically evaluated contrast limited adaptive histogram equalization with varying clip limits for preprocessing chest X-ray images,demonstrating its effectiveness in enhancing feature visibility for diagnostic accuracy.Employing a comprehensive dataset of 5,863 X-ray images(1,583 pneumonia-negative,4,280 pneumonia-positive)collected from multiple healthcare facilities,we conducted a comparative analysis of transfer learning with pre-trained models including ResNet50v2,VGG-19,and MobileNetV2.Statistical validation was performed through 5-fold cross-validation.Results:Our results show that the contrast limited adaptive histogram equalization-enhanced approach with ResNet50v2 achieves 93.40%accuracy,outperforming VGG-19(84.90%)and MobileNetV2(89.70%).Statistical validation confirms the significance of these improvements(P<0.01).The development and optimization resulted in a lightweight mobile application(74 KB)providing rapid diagnostic support(1-2 s response time).Conclusion:The proposed approach demonstrates practical applicability in resource-constrained settings,balancing diagnostic accuracy with deployment efficiency,and offers a viable solution for computer-aided pneumonia diagnosis in areas with limited medical expertise.展开更多
Early detection of Forest and Land Fires(FLF)is essential to prevent the rapid spread of fire as well as minimize environmental damage.However,accurate detection under real-world conditions,such as low light,haze,and ...Early detection of Forest and Land Fires(FLF)is essential to prevent the rapid spread of fire as well as minimize environmental damage.However,accurate detection under real-world conditions,such as low light,haze,and complex backgrounds,remains a challenge for computer vision systems.This study evaluates the impact of three image enhancement techniques—Histogram Equalization(HE),Contrast Limited Adaptive Histogram Equalization(CLAHE),and a hybrid method called DBST-LCM CLAHE—on the performance of the YOLOv11 object detection model in identifying fires and smoke.The D-Fire dataset,consisting of 21,527 annotated images captured under diverse environmental scenarios and illumination levels,was used to train and evaluate the model.Each enhancement method was applied to the dataset before training.Model performance was assessed using multiple metrics,including Precision,Recall,mean Average Precision at 50%IoU(mAP50),F1-score,and visual inspection through bounding box results.Experimental results show that all three enhancement techniques improved detection performance.HE yielded the highest mAP50 score of 0.771,along with a balanced precision of 0.784 and recall of 0.703,demonstrating strong generalization across different conditions.DBST-LCM CLAHE achieved the highest Precision score of 79%,effectively reducing false positives,particularly in scenes with dispersed smoke or complex textures.CLAHE,with slightly lower overall metrics,contributed to improved local feature detection.Each technique showed distinct advantages:HE enhanced global contrast;CLAHE improved local structure visibility;and DBST-LCM CLAHE provided an optimal balance through dynamic block sizing and local contrast preservation.These results underline the importance of selecting preprocessing methods according to detection priorities,such as minimizing false alarms or maximizing completeness.This research does not propose a new model architecture but rather benchmarks a recent lightweight detector,YOLOv11,combined with image enhancement strategies for practical deployment in FLF monitoring.The findings support the integration of preprocessing techniques to improve detection accuracy,offering a foundation for real-time FLF detection systems on edge devices or drones,particularly in regions like Indonesia.展开更多
AIM:To find the effective contrast enhancement method on retinal images for effective segmentation of retinal features.METHODS:A novel image preprocessing method that used neighbourhood-based improved contrast limited...AIM:To find the effective contrast enhancement method on retinal images for effective segmentation of retinal features.METHODS:A novel image preprocessing method that used neighbourhood-based improved contrast limited adaptive histogram equalization(NICLAHE)to improve retinal image contrast was suggested to aid in the accurate identification of retinal disorders and improve the visibility of fine retinal structures.Additionally,a minimal-order filter was applied to effectively denoise the images without compromising important retinal structures.The novel NICLAHE algorithm was inspired by the classical CLAHE algorithm,but enhanced it by selecting the clip limits and tile sized in a dynamical manner relative to the pixel values in an image as opposed to using fixed values.It was evaluated on the Drive and high-resolution fundus(HRF)datasets on conventional quality measures.RESULTS:The new proposed preprocessing technique was applied to two retinal image databases,Drive and HRF,with four quality metrics being,root mean square error(RMSE),peak signal to noise ratio(PSNR),root mean square contrast(RMSC),and overall contrast.The technique performed superiorly on both the data sets as compared to the traditional enhancement methods.In order to assess the compatibility of the method with automated diagnosis,a deep learning framework named ResNet was applied in the segmentation of retinal blood vessels.Sensitivity,specificity,precision and accuracy were used to analyse the performance.NICLAHE–enhanced images outperformed the traditional techniques on both the datasets with improved accuracy.CONCLUSION:NICLAHE provides better results than traditional methods with less error and improved contrastrelated values.These enhanced images are subsequently measured by sensitivity,specificity,precision,and accuracy,which yield a better result in both datasets.展开更多
Perceptual quality assessment for point cloud is critical for immersive metaverse experience and is a challenging task.Firstly,because point cloud is formed by unstructured 3D points that makes the topology more compl...Perceptual quality assessment for point cloud is critical for immersive metaverse experience and is a challenging task.Firstly,because point cloud is formed by unstructured 3D points that makes the topology more complex.Secondly,the quality impairment generally involves both geometric attributes and color properties,where the measurement of the geometric distortion becomes more complex.We propose a perceptual point cloud quality assessment model that follows the perceptual features of Human Visual System(HVS)and the intrinsic characteristics of the point cloud.The point cloud is first pre-processed to extract the geometric skeleton keypoints with graph filtering-based re-sampling,and local neighboring regions around the geometric skeleton keypoints are constructed by K-Nearest Neighbors(KNN)clustering.For geometric distortion,the Point Feature Histogram(PFH)is extracted as the feature descriptor,and the Earth Mover’s Distance(EMD)between the PFHs of the corresponding local neighboring regions in the reference and the distorted point clouds is calculated as the geometric quality measurement.For color distortion,the statistical moments between the corresponding local neighboring regions are computed as the color quality measurement.Finally,the global perceptual quality assessment model is obtained as the linear weighting aggregation of the geometric and color quality measurement.The experimental results on extensive datasets show that the proposed method achieves the leading performance as compared to the state-of-the-art methods with less computing time.Meanwhile,the experimental results also demonstrate the robustness of the proposed method across various distortion types.The source codes are available at https://github.com/llsurreal919/Point Cloud Quality Assessment.展开更多
Frequency-modulated continuous-wave radar enables the non-contact and privacy-preserving recognition of human behavior.However,the accuracy of behavior recognition is directly influenced by the spatial relationship be...Frequency-modulated continuous-wave radar enables the non-contact and privacy-preserving recognition of human behavior.However,the accuracy of behavior recognition is directly influenced by the spatial relationship between human posture and the radar.To address the issue of low accuracy in behavior recognition when the human body is not directly facing the radar,a method combining local outlier factor with Doppler information is proposed for the correction of multi-classifier recognition results.Initially,the information such as distance,velocity,and micro-Doppler spectrogram of the target is obtained using the fast Fourier transform and histogram of oriented gradients-support vector machine methods,followed by preliminary recognition.Subsequently,Platt scaling is employed to transform recognition results into confidence scores,and finally,the Doppler-local outlier factor method is utilized to calibrate the confidence scores,with the highest confidence classifier result considered as the recognition outcome.Experimental results demonstrate that this approach achieves an average recognition accuracy of 96.23%for comprehensive human behavior recognition in various orientations.展开更多
To improve image quality under low illumination conditions,a novel low-light image enhancement method is proposed in this paper based on multi-illumination estimation and multi-scale fusion(MIMS).Firstly,the illuminat...To improve image quality under low illumination conditions,a novel low-light image enhancement method is proposed in this paper based on multi-illumination estimation and multi-scale fusion(MIMS).Firstly,the illumination is processed by contrast-limited adaptive histogram equalization(CLAHE),adaptive complementary gamma function(ACG),and adaptive detail preserving S-curve(ADPS),respectively,to obtain three components.Then,the fusion-relevant features,exposure,and color contrast are selected as the weight maps.Subsequently,these components and weight maps are fused through multi-scale to generate enhanced illumination.Finally,the enhanced images are obtained by multiplying the enhanced illumination and reflectance.Compared with existing approaches,this proposed method achieves an average increase of 0.81%and 2.89%in the structural similarity index measurement(SSIM)and peak signal-to-noise ratio(PSNR),and a decrease of 6.17%and 32.61%in the natural image quality evaluator(NIQE)and gradient magnitude similarity deviation(GMSD),respectively.展开更多
Eigenstructure-based coherence attributes are efficient and mature techniques for large-scale fracture detection. However, in horizontally bedded and continuous strata, buried fractures in high grayscale value zones a...Eigenstructure-based coherence attributes are efficient and mature techniques for large-scale fracture detection. However, in horizontally bedded and continuous strata, buried fractures in high grayscale value zones are difficult to detect. Furthermore, middleand small-scale fractures in fractured zones where migration image energies are usually not concentrated perfectly are also hard to detect because of the fuzzy, clouded shadows owing to low grayscale values. A new fracture enhancement method combined with histogram equalization is proposed to solve these problems. With this method, the contrast between discontinuities and background in coherence images is increased, linear structures are highlighted by stepwise adjustment of the threshold of the coherence image, and fractures are detected at different scales. Application of the method shows that it can also improve fracture cognition and accuracy.展开更多
In this paper the application of image enhancement techniques to potential field data is briefly described and two improved enhancement methods are introduced. One method is derived from the histogram equalization tec...In this paper the application of image enhancement techniques to potential field data is briefly described and two improved enhancement methods are introduced. One method is derived from the histogram equalization technique and automatically determines the color spectra of geophysical maps. Colors can be properly distributed and visual effects and resolution can be enhanced by the method. The other method is based on the modified Radon transform and gradient calculation and is used to detect and enhance linear features in gravity and magnetic images. The method facilites the detection of line segments in the transform domain. Tests with synthetic images and real data show the methods to be effective in feature enhancement.展开更多
Coherence analysis is a powerful tool in seismic interpretation for imaging geological discontinuities such as faults and fractures. However, subtle faults or fractures of one stratum are difficult to be distinguished...Coherence analysis is a powerful tool in seismic interpretation for imaging geological discontinuities such as faults and fractures. However, subtle faults or fractures of one stratum are difficult to be distinguished on coherence sections （time slices or profiles） due to interferences from adjacent strata, especially these with strong reflectivity. In this paper, we propose a coherence enhancement method which applies local histogram specification （LHS） techniques to enhance subtle faults or fractures in the coherence cubes. Unlike the traditional histogram specification （HS） algorithm, our method processes 3D coherence data without discretization. This method partitions a coherence cube into many sub-blocks and self-adaptively specifies the target distribution in each block based on the whole distribution of the coherence cube. Furthermore, the neighboring blocks are partially overlapped to reduce the edge effect. Applications to real datasets show that the new method enhances the details of subtle faults and fractures noticeably.展开更多
Aim Researching the optimal thieshold of image segmentation. M^ethods An adaptiveimages segmentation method based on the entropy of histogram of gray-level picture and genetic. algorithm (GA) was presental. Results ...Aim Researching the optimal thieshold of image segmentation. M^ethods An adaptiveimages segmentation method based on the entropy of histogram of gray-level picture and genetic. algorithm (GA) was presental. Results In our approach, the segmentation problem was formulated as an optimization problem and the fitness of GA which can efficiently search the segmentation parameter space was regarded as the quality criterion. Conclusion The methodcan be adapted for optimal behold segmentation.展开更多
The integration of the Lab model with the extended histogram of oriented gradients （EHOG） is proposed to improve the accuracy of human appearance matching across disjoint camera views under perturbations such as ill...The integration of the Lab model with the extended histogram of oriented gradients （EHOG） is proposed to improve the accuracy of human appearance matching across disjoint camera views under perturbations such as illumination changes and different viewing angles. For the Lab model that describes the global information of observations, a sorted nearest neighbor clustering method is proposed for color clustering and then a partitioned color matching method is used to calculate the color similarity between observations. The Bhattacharya distance is employed for the textural similarity calculation of the EHOG which describes the local information. The global information, which is robust to different viewing angles and scale changes, describes the observations well. Meanwhile, the use of local information, which is robust to illumination changes, can strengthen the discriminative ability of the method. The integration of global and local information improves the accuracy and robustness of the proposed matching approach. Experiments are carried out indoors, and the results show a high matching accuracy of the proposed method.展开更多
A new regression algorithm of an adaptive reduced relevance vector machine is proposed to estimate the illumination chromaticity of an image for the purpose of color constancy. Within the framework of sparse Bayesian ...A new regression algorithm of an adaptive reduced relevance vector machine is proposed to estimate the illumination chromaticity of an image for the purpose of color constancy. Within the framework of sparse Bayesian learning, the algorithm extends the relevance vector machine by combining global and local kernels adaptively in the form of multiple kernels, and the improved locality preserving projection （LLP） is then applied to reduce the column dimension of the multiple kernel input matrix to achieve less training time. To estimate the illumination chromaticity, the algorithm is trained by fuzzy central values of chromaticity histograms of a set of images and the corresponding illuminants. Experiments with real images indicate that the proposed algorithm performs better than the support vector machine and the relevance vector machine while requiring less training time than the relevance vector machine.展开更多
In practical application,mean shift tracking algorithm is easy to generate tracking drift when the target and the background have similar color distribution.Based on the mean shift algorithm,a kind of background weake...In practical application,mean shift tracking algorithm is easy to generate tracking drift when the target and the background have similar color distribution.Based on the mean shift algorithm,a kind of background weaken weight is proposed in the paper firstly.Combining with the object center weight based on the kernel function,the problem of interference of the similar color background can be solved.And then,a model updating strategy is presented to improve the tracking robustness on the influence of occlusion,illumination,deformation and so on.With the test on the sequence of Tiger,the proposed approach provides better performance than the original mean shift tracking algorithm.展开更多
A solution is proposed for the real-time vehicle verification which is an important problem for numerous on- road vehicle applications. First, based on the vertical symmetry characteristics of vehicle images, a vertic...A solution is proposed for the real-time vehicle verification which is an important problem for numerous on- road vehicle applications. First, based on the vertical symmetry characteristics of vehicle images, a vertical symmetrical histograms of oriented gradients （VS-HOG） descriptor is proposed for extracting the image features. In the classification stage, an extreme learning machine （ELM） is used to improve the real-time performance. Experimental data demonstrate that, compared with other classical methods, the vehicle verification algorithm based on VS-HOG and ELM achieves a better trade-off between cost and performance. The computational cost is reduced by using the algorithm, while keeping the performance loss as low as possible. Furthermore, experimental results further show that the proposed vehicle verification method is suitable for on-road vehicle applications due to its better performance both in efficiency and accuracy.展开更多
基金National Key Research and Development Program of China(2022YFB2804401)。
文摘An in-pixel histogramming time-to-digital converter(hTDC)based on octonary search and 4-tap phase detection is presented,aiming to improve frame rate while ensuring high precicion.The proposed hTDC is a 12-bit two-step converter consisting of a 6-bit coarse quantization and a 6-bit fine quantization,which supports a time resolution of 120 ps and multiphoton counting up to 2 GHz without a GHz reference frequency.The proposed hTDC is designed in 0.11μm CMOS process with an area consumption of 6900μm^(2).The data from a behavioral-level model is imported into the designed hTDC circuit for simulation verification.The post-simulation results show that the proposed hTDC achieves 0.8%depth precision in 9 m range for short-range system design specifications and 0.2%depth precision in 48 m range for long-range system design specifications.Under 30×10^(3) lux background light conditions,the proposed hTDC can be used for SPAD-based flash LiDAR sensor to achieve a frame rate to 40 fps with 200 ps resolution in 9 m range.
基金National Natural Science Foundation of China(62071147)。
文摘Ground penetrating radar(GPR),as a fast,efficient,and non-destructive detection device,holds great potential for the detection of shallow subsurface environments,such as urban road subsurface monitoring.However,the interpretation of GPR echo images often relies on manual recognition by experienced engineers.In order to address the automatic interpretation of cavity targets in GPR echo images,a recognition-algorithm based on Gaussian mixed model-hidden Markov model(GMM-HMM)is proposed,which can recognize three dimensional(3D)underground voids automatically.First,energy detection on the echo images is performed,whereby the data is preprocessed and pre-filtered.Then,edge histogram descriptor(EHD),histogram of oriented gradient(HOG),and Log-Gabor filters are used to extract features from the images.The traditional method can only be applied to 2D images and pre-processing is required for C-scan images.Finally,the aggregated features are fed into the GMM-HMM for classification and compared with two other methods,long short-term memory(LSTM)and gate recurrent unit(GRU).By testing on a simulated dataset,an accuracy rate of 90%is obtained,demonstrating the effectiveness and efficiency of our proposed method.
文摘Skin cancer,a severe health threat,can spread rapidly if undetected.Therefore,early detection can lead to an advanced and efficient diagnosis,thus reducing mortality.Unsupervised classification techniques analyse extensive skin image datasets,identifying patterns and anomalies without prior labelling,facilitating early detection and effective diagnosis and potentially saving lives.In this study,the authors aim to explore the potential of unsupervised learning methods in classifying different types of skin lesions in dermatoscopic images.The authors aim to bridge the gap in dermatological research by introducing innovative techniques that enhance image quality and improve feature extraction.To achieve this,enhanced super-resolution generative adversarial networks(ESRGAN)was fine-tuned to strengthen the resolution of skin lesion images,making critical features more visible.The authors extracted histogram features to capture essential colour characteristics and used the Davies-Bouldin index and silhouette score to determine optimal clusters.Fine-tuned k-means clustering with Euclidean distance in the histogram feature space achieved 87.77% and 90.5% test accuracies on the ISIC2019 and HAM10000 datasets,respectively.The unsupervised approach effectively categorises skin lesions,indicating that unsupervised learning can significantly advance dermatology by enabling early detection and classification without extensive manual annotation.
文摘Esophageal cancer is one of the most difficult cancers to treat since it is often at an advanced stage at the time of symptom presentation.For locally advanced esophageal cancer,treatment options include multidisciplinary treatment such as surgery or definitive chemoradiotherapy.Surgery has a high local control rate because it involves excision of the cancer along with the surrounding organs;however,it is still highly invasive,although advances in surgery have reduced the burden on patients.On the other hand,chemoradiotherapy may also be applicable in cases in which surgery is inoperable owing to complications or distant lymph node metastasis.However,chemoradiotherapy using X-ray irradiation can cause late toxicities,including those to the heart.Proton beam therapy is widely used to treat esophageal cancer because of its characteristics,and some comparisons between proton beam therapy and X-ray therapy or surgery have recently been reported.This review discusses the role of proton beam therapy in esophageal cancer in comparison to X-ray therapy and surgery.
文摘Background:Pneumonia remains a critical global health challenge,manifesting as a severe respiratory infection caused by viruses,bacteria,and fungi.Early detection is paramount for effective treatment,potentially reducing mortality rates and optimizing healthcare resource allocation.Despite the importance of chest X-ray diagnosis,image analysis presents significant challenges,particularly in regions with limited medical expertise.This study addresses these challenges by proposing a computer-aided diagnosis system leveraging targeted image preprocessing and optimized deep learning techniques.Methods:We systematically evaluated contrast limited adaptive histogram equalization with varying clip limits for preprocessing chest X-ray images,demonstrating its effectiveness in enhancing feature visibility for diagnostic accuracy.Employing a comprehensive dataset of 5,863 X-ray images(1,583 pneumonia-negative,4,280 pneumonia-positive)collected from multiple healthcare facilities,we conducted a comparative analysis of transfer learning with pre-trained models including ResNet50v2,VGG-19,and MobileNetV2.Statistical validation was performed through 5-fold cross-validation.Results:Our results show that the contrast limited adaptive histogram equalization-enhanced approach with ResNet50v2 achieves 93.40%accuracy,outperforming VGG-19(84.90%)and MobileNetV2(89.70%).Statistical validation confirms the significance of these improvements(P<0.01).The development and optimization resulted in a lightweight mobile application(74 KB)providing rapid diagnostic support(1-2 s response time).Conclusion:The proposed approach demonstrates practical applicability in resource-constrained settings,balancing diagnostic accuracy with deployment efficiency,and offers a viable solution for computer-aided pneumonia diagnosis in areas with limited medical expertise.
基金funded by the Directorate of Research,Technology,and Community Service,Ministry of Higher Education,Science,and Technology of the Republic of Indonesia the Regular Fundamental Research scheme,with grant numbers 001/LL6/PL/AL.04/2025,011/SPK-PFR/RIK/05/2025.
文摘Early detection of Forest and Land Fires(FLF)is essential to prevent the rapid spread of fire as well as minimize environmental damage.However,accurate detection under real-world conditions,such as low light,haze,and complex backgrounds,remains a challenge for computer vision systems.This study evaluates the impact of three image enhancement techniques—Histogram Equalization(HE),Contrast Limited Adaptive Histogram Equalization(CLAHE),and a hybrid method called DBST-LCM CLAHE—on the performance of the YOLOv11 object detection model in identifying fires and smoke.The D-Fire dataset,consisting of 21,527 annotated images captured under diverse environmental scenarios and illumination levels,was used to train and evaluate the model.Each enhancement method was applied to the dataset before training.Model performance was assessed using multiple metrics,including Precision,Recall,mean Average Precision at 50%IoU(mAP50),F1-score,and visual inspection through bounding box results.Experimental results show that all three enhancement techniques improved detection performance.HE yielded the highest mAP50 score of 0.771,along with a balanced precision of 0.784 and recall of 0.703,demonstrating strong generalization across different conditions.DBST-LCM CLAHE achieved the highest Precision score of 79%,effectively reducing false positives,particularly in scenes with dispersed smoke or complex textures.CLAHE,with slightly lower overall metrics,contributed to improved local feature detection.Each technique showed distinct advantages:HE enhanced global contrast;CLAHE improved local structure visibility;and DBST-LCM CLAHE provided an optimal balance through dynamic block sizing and local contrast preservation.These results underline the importance of selecting preprocessing methods according to detection priorities,such as minimizing false alarms or maximizing completeness.This research does not propose a new model architecture but rather benchmarks a recent lightweight detector,YOLOv11,combined with image enhancement strategies for practical deployment in FLF monitoring.The findings support the integration of preprocessing techniques to improve detection accuracy,offering a foundation for real-time FLF detection systems on edge devices or drones,particularly in regions like Indonesia.
文摘AIM:To find the effective contrast enhancement method on retinal images for effective segmentation of retinal features.METHODS:A novel image preprocessing method that used neighbourhood-based improved contrast limited adaptive histogram equalization(NICLAHE)to improve retinal image contrast was suggested to aid in the accurate identification of retinal disorders and improve the visibility of fine retinal structures.Additionally,a minimal-order filter was applied to effectively denoise the images without compromising important retinal structures.The novel NICLAHE algorithm was inspired by the classical CLAHE algorithm,but enhanced it by selecting the clip limits and tile sized in a dynamical manner relative to the pixel values in an image as opposed to using fixed values.It was evaluated on the Drive and high-resolution fundus(HRF)datasets on conventional quality measures.RESULTS:The new proposed preprocessing technique was applied to two retinal image databases,Drive and HRF,with four quality metrics being,root mean square error(RMSE),peak signal to noise ratio(PSNR),root mean square contrast(RMSC),and overall contrast.The technique performed superiorly on both the data sets as compared to the traditional enhancement methods.In order to assess the compatibility of the method with automated diagnosis,a deep learning framework named ResNet was applied in the segmentation of retinal blood vessels.Sensitivity,specificity,precision and accuracy were used to analyse the performance.NICLAHE–enhanced images outperformed the traditional techniques on both the datasets with improved accuracy.CONCLUSION:NICLAHE provides better results than traditional methods with less error and improved contrastrelated values.These enhanced images are subsequently measured by sensitivity,specificity,precision,and accuracy,which yield a better result in both datasets.
基金supported in part by the National Natural Science Foundation of China under Grant(62171257,U22B2001,U19A2052,62020106011,62061015)in part by the Natural Science Foundation of Chongqing under Grant(2023NSCQMSX2930)+1 种基金in part by the Youth Innovation Group Support Program of ICE Discipline of CQUPT under Grant(SCIE-QN-2022-05)in part by the Graduate Scientifc Research and Innovation Project of Chongqing under Grant(CYS22469).
文摘Perceptual quality assessment for point cloud is critical for immersive metaverse experience and is a challenging task.Firstly,because point cloud is formed by unstructured 3D points that makes the topology more complex.Secondly,the quality impairment generally involves both geometric attributes and color properties,where the measurement of the geometric distortion becomes more complex.We propose a perceptual point cloud quality assessment model that follows the perceptual features of Human Visual System(HVS)and the intrinsic characteristics of the point cloud.The point cloud is first pre-processed to extract the geometric skeleton keypoints with graph filtering-based re-sampling,and local neighboring regions around the geometric skeleton keypoints are constructed by K-Nearest Neighbors(KNN)clustering.For geometric distortion,the Point Feature Histogram(PFH)is extracted as the feature descriptor,and the Earth Mover’s Distance(EMD)between the PFHs of the corresponding local neighboring regions in the reference and the distorted point clouds is calculated as the geometric quality measurement.For color distortion,the statistical moments between the corresponding local neighboring regions are computed as the color quality measurement.Finally,the global perceptual quality assessment model is obtained as the linear weighting aggregation of the geometric and color quality measurement.The experimental results on extensive datasets show that the proposed method achieves the leading performance as compared to the state-of-the-art methods with less computing time.Meanwhile,the experimental results also demonstrate the robustness of the proposed method across various distortion types.The source codes are available at https://github.com/llsurreal919/Point Cloud Quality Assessment.
基金the National Key Research and Development Program of China(No.2022YFC3601400)。
文摘Frequency-modulated continuous-wave radar enables the non-contact and privacy-preserving recognition of human behavior.However,the accuracy of behavior recognition is directly influenced by the spatial relationship between human posture and the radar.To address the issue of low accuracy in behavior recognition when the human body is not directly facing the radar,a method combining local outlier factor with Doppler information is proposed for the correction of multi-classifier recognition results.Initially,the information such as distance,velocity,and micro-Doppler spectrogram of the target is obtained using the fast Fourier transform and histogram of oriented gradients-support vector machine methods,followed by preliminary recognition.Subsequently,Platt scaling is employed to transform recognition results into confidence scores,and finally,the Doppler-local outlier factor method is utilized to calibrate the confidence scores,with the highest confidence classifier result considered as the recognition outcome.Experimental results demonstrate that this approach achieves an average recognition accuracy of 96.23%for comprehensive human behavior recognition in various orientations.
基金supported by the National Key R&D Program of China(No.2022YFB3205101)NSAF(No.U2230116)。
文摘To improve image quality under low illumination conditions,a novel low-light image enhancement method is proposed in this paper based on multi-illumination estimation and multi-scale fusion(MIMS).Firstly,the illumination is processed by contrast-limited adaptive histogram equalization(CLAHE),adaptive complementary gamma function(ACG),and adaptive detail preserving S-curve(ADPS),respectively,to obtain three components.Then,the fusion-relevant features,exposure,and color contrast are selected as the weight maps.Subsequently,these components and weight maps are fused through multi-scale to generate enhanced illumination.Finally,the enhanced images are obtained by multiplying the enhanced illumination and reflectance.Compared with existing approaches,this proposed method achieves an average increase of 0.81%and 2.89%in the structural similarity index measurement(SSIM)and peak signal-to-noise ratio(PSNR),and a decrease of 6.17%and 32.61%in the natural image quality evaluator(NIQE)and gradient magnitude similarity deviation(GMSD),respectively.
基金sponsored by the National Science&Technology Major Special Project(Grant No.2011ZX05025-001-04)
文摘Eigenstructure-based coherence attributes are efficient and mature techniques for large-scale fracture detection. However, in horizontally bedded and continuous strata, buried fractures in high grayscale value zones are difficult to detect. Furthermore, middleand small-scale fractures in fractured zones where migration image energies are usually not concentrated perfectly are also hard to detect because of the fuzzy, clouded shadows owing to low grayscale values. A new fracture enhancement method combined with histogram equalization is proposed to solve these problems. With this method, the contrast between discontinuities and background in coherence images is increased, linear structures are highlighted by stepwise adjustment of the threshold of the coherence image, and fractures are detected at different scales. Application of the method shows that it can also improve fracture cognition and accuracy.
基金This work is supported by the research project (grant No. G20000467) of the Institute of Geology and Geophysics, CAS and bythe China Postdoctoral Science Foundation (No. 2004036083).
文摘In this paper the application of image enhancement techniques to potential field data is briefly described and two improved enhancement methods are introduced. One method is derived from the histogram equalization technique and automatically determines the color spectra of geophysical maps. Colors can be properly distributed and visual effects and resolution can be enhanced by the method. The other method is based on the modified Radon transform and gradient calculation and is used to detect and enhance linear features in gravity and magnetic images. The method facilites the detection of line segments in the transform domain. Tests with synthetic images and real data show the methods to be effective in feature enhancement.
基金sponsored by Important National Science and Technology Specific Projects of China (Grant No.2008ZX05023-005-011 and No. 2008ZX05040-003)the National 973 Program of China (Grant No. 2006CB202208)
文摘Coherence analysis is a powerful tool in seismic interpretation for imaging geological discontinuities such as faults and fractures. However, subtle faults or fractures of one stratum are difficult to be distinguished on coherence sections （time slices or profiles） due to interferences from adjacent strata, especially these with strong reflectivity. In this paper, we propose a coherence enhancement method which applies local histogram specification （LHS） techniques to enhance subtle faults or fractures in the coherence cubes. Unlike the traditional histogram specification （HS） algorithm, our method processes 3D coherence data without discretization. This method partitions a coherence cube into many sub-blocks and self-adaptively specifies the target distribution in each block based on the whole distribution of the coherence cube. Furthermore, the neighboring blocks are partially overlapped to reduce the edge effect. Applications to real datasets show that the new method enhances the details of subtle faults and fractures noticeably.
文摘Aim Researching the optimal thieshold of image segmentation. M^ethods An adaptiveimages segmentation method based on the entropy of histogram of gray-level picture and genetic. algorithm (GA) was presental. Results In our approach, the segmentation problem was formulated as an optimization problem and the fitness of GA which can efficiently search the segmentation parameter space was regarded as the quality criterion. Conclusion The methodcan be adapted for optimal behold segmentation.
基金The National Natural Science Foundation of China(No.60972001)the Science and Technology Plan of Suzhou City(No.SG201076)
文摘The integration of the Lab model with the extended histogram of oriented gradients （EHOG） is proposed to improve the accuracy of human appearance matching across disjoint camera views under perturbations such as illumination changes and different viewing angles. For the Lab model that describes the global information of observations, a sorted nearest neighbor clustering method is proposed for color clustering and then a partitioned color matching method is used to calculate the color similarity between observations. The Bhattacharya distance is employed for the textural similarity calculation of the EHOG which describes the local information. The global information, which is robust to different viewing angles and scale changes, describes the observations well. Meanwhile, the use of local information, which is robust to illumination changes, can strengthen the discriminative ability of the method. The integration of global and local information improves the accuracy and robustness of the proposed matching approach. Experiments are carried out indoors, and the results show a high matching accuracy of the proposed method.
基金The National Natural Science Foundation of China(No60573139)the Innovation Foundation of Xidian University forGraduates (No05008)
文摘A new regression algorithm of an adaptive reduced relevance vector machine is proposed to estimate the illumination chromaticity of an image for the purpose of color constancy. Within the framework of sparse Bayesian learning, the algorithm extends the relevance vector machine by combining global and local kernels adaptively in the form of multiple kernels, and the improved locality preserving projection （LLP） is then applied to reduce the column dimension of the multiple kernel input matrix to achieve less training time. To estimate the illumination chromaticity, the algorithm is trained by fuzzy central values of chromaticity histograms of a set of images and the corresponding illuminants. Experiments with real images indicate that the proposed algorithm performs better than the support vector machine and the relevance vector machine while requiring less training time than the relevance vector machine.
基金National Natural Science Foundation of China(No.61201412)
文摘In practical application,mean shift tracking algorithm is easy to generate tracking drift when the target and the background have similar color distribution.Based on the mean shift algorithm,a kind of background weaken weight is proposed in the paper firstly.Combining with the object center weight based on the kernel function,the problem of interference of the similar color background can be solved.And then,a model updating strategy is presented to improve the tracking robustness on the influence of occlusion,illumination,deformation and so on.With the test on the sequence of Tiger,the proposed approach provides better performance than the original mean shift tracking algorithm.
基金The National Natural Science Foundation of China(No.61203237)the Natural Science Foundation of Zhejiang Province(No.LQ12F03016)the China Postdoctoral Science Foundation(No.2011M500836)
文摘A solution is proposed for the real-time vehicle verification which is an important problem for numerous on- road vehicle applications. First, based on the vertical symmetry characteristics of vehicle images, a vertical symmetrical histograms of oriented gradients （VS-HOG） descriptor is proposed for extracting the image features. In the classification stage, an extreme learning machine （ELM） is used to improve the real-time performance. Experimental data demonstrate that, compared with other classical methods, the vehicle verification algorithm based on VS-HOG and ELM achieves a better trade-off between cost and performance. The computational cost is reduced by using the algorithm, while keeping the performance loss as low as possible. Furthermore, experimental results further show that the proposed vehicle verification method is suitable for on-road vehicle applications due to its better performance both in efficiency and accuracy.
