Fisheye cameras offer a significantly larger field of view compared to conventional cameras,making them valuable tools in the field of computer vision.However,their unique optical characteristics often lead to image d...Fisheye cameras offer a significantly larger field of view compared to conventional cameras,making them valuable tools in the field of computer vision.However,their unique optical characteristics often lead to image distortions,which pose challenges for object detection tasks.To address this issue,we propose Yolo-CaSKA(Yolo with Contrastive Learning and Selective Kernel Attention),a novel training method that enhances object detection on fisheye camera images.The standard image and the corresponding distorted fisheye image pairs are used as positive samples,and the rest of the image pairs are used as negative samples,which are guided by contrastive learning to help the distorted images find the feature vectors of the corresponding normal images,to improve the detection accuracy.Additionally,we incorporate the Selective Kernel(SK)attention module to focus on regions prone to false detections,such as image edges and blind spots.Finally,the mAP_(50) on the augmented KITTI dataset is improved by 5.5% over the original Yolov8,while the mAP_(50) on the WoodScape dataset is improved by 2.6% compared to OmniDet.The results demonstrate the performance of our proposed model for object detection on fisheye images.展开更多
Top-view fisheye cameras are widely used in personnel surveillance for their broad field of view,but their unique imaging characteristics pose challenges like distortion,complex scenes,scale variations,and small objec...Top-view fisheye cameras are widely used in personnel surveillance for their broad field of view,but their unique imaging characteristics pose challenges like distortion,complex scenes,scale variations,and small objects near image edges.To tackle these,we proposed peripheral focus you only look once(PF-YOLO),an enhanced YOLOv8n-based method.Firstly,we introduced a cutting-patch data augmentation strategy to mitigate the problem of insufficient small-object samples in various scenes.Secondly,to enhance the model's focus on small objects near the edges,we designed the peripheral focus loss,which uses dynamic focus coefficients to provide greater gradient gains for these objects,improving their regression accuracy.Finally,we designed the three dimensional(3D)spatial-channel coordinate attention C2f module,enhancing spatial and channel perception,suppressing noise,and improving personnel detection.Experimental results demonstrate that PF-YOLO achieves strong performance on the challenging events for person detection from overhead fisheye images(CEPDTOF)and in-the-wild events for people detection and tracking from overhead fisheye cameras(WEPDTOF)datasets.Compared to the original YOLOv8n model,PFYOLO achieves improvements on CEPDTOF with increases of 2.1%,1.7%and 2.9%in mean average precision 50(mAP 50),mAP 50-95,and tively.On WEPDTOF,PF-YOLO achieves substantial improvements with increases of 31.4%,14.9%,61.1%and 21.0%in 91.2%and 57.2%,respectively.展开更多
Forest canopy in a deciduous forest has significant sheltering effects on the sub-canopy solar radiation,significantly influencing the energy balance of snow and permafrost beneath the forest and their spatial distrib...Forest canopy in a deciduous forest has significant sheltering effects on the sub-canopy solar radiation,significantly influencing the energy balance of snow and permafrost beneath the forest and their spatial distribution.This study employs a digital camera mounted with a fisheye lens to acquire photographs at various times in a growth cycle of the forest canopy at three selected sites in a deciduous forest near the Greater Khingan Mountains Forest Ecological Station,Northeast China.The vegetation types and conditions at the selected sites include P1 in Ledum-Claopodium-L.dahurica,P2 in Carex tato-L.dahurica,and P3 in Betula fruticosa-L.dahurica.After necessary image processing,these photographs were used to identify the canopy structure and its impacts on the sub-canopy solar radiation.Results show that fisheye photographs can successfully capture the forest canopy structure and are useful in estimating the sub-canopy solar radiation.The order of sheltering effects from the largest to the smallest on sub-canopy solar radiation at three selected sites is P3,P1,and P2,highly depending on the canopy density.Then sub-canopy solar radiation was calculated using fisheye photographs and an algorithm validated by in-situ observed solar radiation beneath the canopy at P1 and P3.The results are reasonable,although the accuracy seems compromised due to the mismatch of conditions for calculation and observation.Results also show that the mean annual solar radiation above the canopy was about 148.3 W/m2 in 2018,and the mean annual solar radiation values beneath the canopy were about 90.0,123.8,and 61.0 W/m2 at P1,P2,and P3,with only 60%,84%,and 42%of the total solar radiation penetrating through the canopy,respectively.Even in winter,when the trees are leafless,the canopy sheltering effects cannot be ignored in dense forests.Despite the limitations,fisheye photographs and related algorithms are useful in investigating the forest canopy structure and its impacts on sub-canopy solar radiation.展开更多
This paper presents a low-cost remote vision system for use in unmanned aircraft that provide a first person view （FPV） to vehicle operators in real-time. The system does not require a traditional electromechanical ...This paper presents a low-cost remote vision system for use in unmanned aircraft that provide a first person view （FPV） to vehicle operators in real-time. The system does not require a traditional electromechanical gimbal setup. Instead, the system uses a wide-angle （fisheye） lens and a video camera setup that is fixed on the vehicle and captures the full viewing area as seen from the cockpit in each video frame. Video is transmitted to a ground station wirelessly. On the ground, the pilot is outfitted with virtual reality goggles with integrated attitude and heading sensors. The received video is recertified and cropped by the ground station to provide the goggles with the appropriate view based on head orientation. Compared to traditional electromechanical setups, the presented system features reduced weight, reduced video lag, lower power consumption, and reduced drag on the airborne vehicle in addition to requiring only a unidirectional downlink. The video processing is preformed on the ground, further reducing computational resources and bandwidth requirements. These advantages, in conjunction with the advancement in miniature optical sensors and lenses, make the proposed approach a viable option for miniature remotely controlled vehicles. The system was successfully implemented and tested using an R/C airplane.展开更多
A design for a fisheye optical system is modified in order to enable IR vision in addition to the visible and UV lighting. The proposed modification goes to replace the optical material of the movable lens by another ...A design for a fisheye optical system is modified in order to enable IR vision in addition to the visible and UV lighting. The proposed modification goes to replace the optical material of the movable lens by another with less thermal dispersion. The choice of appropriate materials lead to a good focus appeared on the retina for a wide spectral range includes UV, visible, and near IR lighting. Then, the performance of the modified design is verified through some optical measures for imaging quality determinations. These optical measures are determined with the aid of Zemax software, which also used for testing performance of the modified fisheye optical system. The analysis mainly focuses on the energy distribution in the light spot on the focal surface. The results show that the modified design of the fisheye is acceptable.展开更多
随着单反相机跨入数码时代,APS-C画幅的数码单反开始普及,同时,鱼眼镜头也开始陷入危机,1.5x-1.7x的焦距转换系数会让原本十分"宽广"的视角变得"狭窄",原有的视觉冲击力也大打折扣。于是,专为APS- C画幅数码单反相...随着单反相机跨入数码时代,APS-C画幅的数码单反开始普及,同时,鱼眼镜头也开始陷入危机,1.5x-1.7x的焦距转换系数会让原本十分"宽广"的视角变得"狭窄",原有的视觉冲击力也大打折扣。于是,专为APS- C画幅数码单反相机设计的数码专用鱼眼镜头应运而生。我们要体验的适马4.5mm f/2.8 EX DC Circular Fisheye HSM正是这样一款数码专用鱼眼镜头。展开更多
一直以来,鱼眼镜头以其超宽的视角和独特的视觉效果,引来无数摄友的目光;其"足迹"更是遍布风光、新闻、体育、纪实等各个领域,成为很多摄影师必不可少的利器。本期我们要体验的就是适马10mm f/2.8 EX DC Fisheye HSM这款数码...一直以来,鱼眼镜头以其超宽的视角和独特的视觉效果,引来无数摄友的目光;其"足迹"更是遍布风光、新闻、体育、纪实等各个领域,成为很多摄影师必不可少的利器。本期我们要体验的就是适马10mm f/2.8 EX DC Fisheye HSM这款数码专用鱼眼镜头,测试机身为佳能EOS 20D。展开更多
In complicated urban environments,Global Navigation Satellite System(GNSS)signals are frequently affected by building reflection or refraction,resulting in Non-Line-of-Sight(NLOS)errors.In severe cases,NLOS errors can...In complicated urban environments,Global Navigation Satellite System(GNSS)signals are frequently affected by building reflection or refraction,resulting in Non-Line-of-Sight(NLOS)errors.In severe cases,NLOS errors can cause a ranging error of hundreds of meters,which has a substantial impact on the precision and dependability of GNSS positioning.To address this problem,we propose a reliable NLOS error identification method based on the Light Gradient Boosting Machine(LightGBM),which is driven by multiple features of GNSS signals.The sample data are first labeled using a fisheye camera to classify the signals from visible satellites as Line-of-Sight(LOS)or NLOS signals.We then analyzed the sample data to determine the correlation among multiple features,such as the signal-to-noise ratio,elevation angle,pseudorange consistency,phase consistency,Code Minus Carrier,and Multi-Path combined observations.Finally,we introduce the LightGBM model to establish an effective correlation between signal features and satellite visibility and adopt a multifeature-driven scheme to achieve reliable identification of NLOSs.The test results show that the proposed method is superior to other methods such as Extreme Gradient Boosting(XGBoost),in terms of accuracy and usability.The model demonstrates a potential classification accuracy of approximately 90%with minimal time consumption.Furthermore,the Standard Point Positioning results after excluding NLOSs show the Root Mean Squares are improved by 47.82%,56.68%,and 36.68%in the east,north,and up directions,respectively,and the overall positioning performance is significantly improved.展开更多
Analysis of forest canopy hemisphere images is one of the most important methods for measuring forest canopy structure parameters. In this study, our main focus was on using circular image region segmentation, which i...Analysis of forest canopy hemisphere images is one of the most important methods for measuring forest canopy structure parameters. In this study, our main focus was on using circular image region segmentation, which is the basis of forest canopy hemispherical photography. The boundary of a forest canopy hemisphere image was analyzed via histogram, rectangle, and Fourier descriptors. The image boundary characteristics were defined and obtained based on the following：（1） an edge model that contains three parts, i.e., step, ramp, and roof;（2） boundary points of discontinuity;（3） an edge that has a linear distribution of scattering points. On this basis, we proposed a segmentation method for the circular region in a forest canopy hemisphere image, fitting the circular boundary and computing the center and radius by the least squares method. The method was unrelated to the parameters of the image acquisition device. Hence, this study lays a foundation for automatically adjusting the parameters of high-performance image acquisition devices used in forest canopy hemispherical photography.展开更多
The spatial distribution of power facilities is uneven,thereby making the topology of geographical wiring diagrams(GWDs)based on the actual coordinates unclear.A single-line diagram has the advantage of a clear topolo...The spatial distribution of power facilities is uneven,thereby making the topology of geographical wiring diagrams(GWDs)based on the actual coordinates unclear.A single-line diagram has the advantage of a clear topology but it lacks spatial locations.A GWD has the advantage of accurate spatial locations but it lacks a clear topology.Visualizing distribution networks for planning requires both features.We proposed a new planning-oriented method for optimizing the visualization of distribution networks.From the global perspective,we proposed an improved force-directed(FD)algorithm by introducing a space restriction strategy and node–edge repulsion strategy to promote the expansion of the distance between distribution facilities within a limited buffer.We then constructed the constrained Delaunay triangulation to identify the compact districts(CDs)and used a genetic algorithm to optimize the parameters for the improved FD algorithm.A novel visualization evaluation indicator was also proposed for quantitatively assessing the visualizations.From a local perspective,the fisheye algorithm was used to optimize the CDs to further improve the visualization of the distribution network.We verified the proposed methods with real-world data.We used limited spatial displacement in exchange for maximum topology clarity to balance the accurate spatial location and topology clarity.展开更多
文摘Fisheye cameras offer a significantly larger field of view compared to conventional cameras,making them valuable tools in the field of computer vision.However,their unique optical characteristics often lead to image distortions,which pose challenges for object detection tasks.To address this issue,we propose Yolo-CaSKA(Yolo with Contrastive Learning and Selective Kernel Attention),a novel training method that enhances object detection on fisheye camera images.The standard image and the corresponding distorted fisheye image pairs are used as positive samples,and the rest of the image pairs are used as negative samples,which are guided by contrastive learning to help the distorted images find the feature vectors of the corresponding normal images,to improve the detection accuracy.Additionally,we incorporate the Selective Kernel(SK)attention module to focus on regions prone to false detections,such as image edges and blind spots.Finally,the mAP_(50) on the augmented KITTI dataset is improved by 5.5% over the original Yolov8,while the mAP_(50) on the WoodScape dataset is improved by 2.6% compared to OmniDet.The results demonstrate the performance of our proposed model for object detection on fisheye images.
基金supported by National Natural Science Foundation of China(Nos.62171042,62102033,U24A20331)the R&D Program of Beijing Municipal Education Commission(No.KZ202211417048)+2 种基金the Project of Construction and Support for High-Level Innovative Teams of Beijing Municipal Institutions(No.BPHR20220121)Beijing Natural Science Foundation(Nos.4232026,4242020)the Academic Research Projects of Beijing Union University(Nos.ZKZD202302,ZK20202403)。
文摘Top-view fisheye cameras are widely used in personnel surveillance for their broad field of view,but their unique imaging characteristics pose challenges like distortion,complex scenes,scale variations,and small objects near image edges.To tackle these,we proposed peripheral focus you only look once(PF-YOLO),an enhanced YOLOv8n-based method.Firstly,we introduced a cutting-patch data augmentation strategy to mitigate the problem of insufficient small-object samples in various scenes.Secondly,to enhance the model's focus on small objects near the edges,we designed the peripheral focus loss,which uses dynamic focus coefficients to provide greater gradient gains for these objects,improving their regression accuracy.Finally,we designed the three dimensional(3D)spatial-channel coordinate attention C2f module,enhancing spatial and channel perception,suppressing noise,and improving personnel detection.Experimental results demonstrate that PF-YOLO achieves strong performance on the challenging events for person detection from overhead fisheye images(CEPDTOF)and in-the-wild events for people detection and tracking from overhead fisheye cameras(WEPDTOF)datasets.Compared to the original YOLOv8n model,PFYOLO achieves improvements on CEPDTOF with increases of 2.1%,1.7%and 2.9%in mean average precision 50(mAP 50),mAP 50-95,and tively.On WEPDTOF,PF-YOLO achieves substantial improvements with increases of 31.4%,14.9%,61.1%and 21.0%in 91.2%and 57.2%,respectively.
基金the National Natural Science Foundation of China(Grant Nos.41971079 and 41671059,41975081).
文摘Forest canopy in a deciduous forest has significant sheltering effects on the sub-canopy solar radiation,significantly influencing the energy balance of snow and permafrost beneath the forest and their spatial distribution.This study employs a digital camera mounted with a fisheye lens to acquire photographs at various times in a growth cycle of the forest canopy at three selected sites in a deciduous forest near the Greater Khingan Mountains Forest Ecological Station,Northeast China.The vegetation types and conditions at the selected sites include P1 in Ledum-Claopodium-L.dahurica,P2 in Carex tato-L.dahurica,and P3 in Betula fruticosa-L.dahurica.After necessary image processing,these photographs were used to identify the canopy structure and its impacts on the sub-canopy solar radiation.Results show that fisheye photographs can successfully capture the forest canopy structure and are useful in estimating the sub-canopy solar radiation.The order of sheltering effects from the largest to the smallest on sub-canopy solar radiation at three selected sites is P3,P1,and P2,highly depending on the canopy density.Then sub-canopy solar radiation was calculated using fisheye photographs and an algorithm validated by in-situ observed solar radiation beneath the canopy at P1 and P3.The results are reasonable,although the accuracy seems compromised due to the mismatch of conditions for calculation and observation.Results also show that the mean annual solar radiation above the canopy was about 148.3 W/m2 in 2018,and the mean annual solar radiation values beneath the canopy were about 90.0,123.8,and 61.0 W/m2 at P1,P2,and P3,with only 60%,84%,and 42%of the total solar radiation penetrating through the canopy,respectively.Even in winter,when the trees are leafless,the canopy sheltering effects cannot be ignored in dense forests.Despite the limitations,fisheye photographs and related algorithms are useful in investigating the forest canopy structure and its impacts on sub-canopy solar radiation.
文摘This paper presents a low-cost remote vision system for use in unmanned aircraft that provide a first person view （FPV） to vehicle operators in real-time. The system does not require a traditional electromechanical gimbal setup. Instead, the system uses a wide-angle （fisheye） lens and a video camera setup that is fixed on the vehicle and captures the full viewing area as seen from the cockpit in each video frame. Video is transmitted to a ground station wirelessly. On the ground, the pilot is outfitted with virtual reality goggles with integrated attitude and heading sensors. The received video is recertified and cropped by the ground station to provide the goggles with the appropriate view based on head orientation. Compared to traditional electromechanical setups, the presented system features reduced weight, reduced video lag, lower power consumption, and reduced drag on the airborne vehicle in addition to requiring only a unidirectional downlink. The video processing is preformed on the ground, further reducing computational resources and bandwidth requirements. These advantages, in conjunction with the advancement in miniature optical sensors and lenses, make the proposed approach a viable option for miniature remotely controlled vehicles. The system was successfully implemented and tested using an R/C airplane.
文摘A design for a fisheye optical system is modified in order to enable IR vision in addition to the visible and UV lighting. The proposed modification goes to replace the optical material of the movable lens by another with less thermal dispersion. The choice of appropriate materials lead to a good focus appeared on the retina for a wide spectral range includes UV, visible, and near IR lighting. Then, the performance of the modified design is verified through some optical measures for imaging quality determinations. These optical measures are determined with the aid of Zemax software, which also used for testing performance of the modified fisheye optical system. The analysis mainly focuses on the energy distribution in the light spot on the focal surface. The results show that the modified design of the fisheye is acceptable.
文摘随着单反相机跨入数码时代,APS-C画幅的数码单反开始普及,同时,鱼眼镜头也开始陷入危机,1.5x-1.7x的焦距转换系数会让原本十分"宽广"的视角变得"狭窄",原有的视觉冲击力也大打折扣。于是,专为APS- C画幅数码单反相机设计的数码专用鱼眼镜头应运而生。我们要体验的适马4.5mm f/2.8 EX DC Circular Fisheye HSM正是这样一款数码专用鱼眼镜头。
基金funded by the National Science Fund for Distinguished Young Scholars of China (Grant No.42425003)the National Natural Science Foundation of China (Grant Nos.42274034,42388102)+2 种基金the Major Program (JD)of Hubei Province (Grant No.2023BAA026)the Special Fund of Hubei Luojia Laboratory (Grant No.2201000038)the Special Fund of Wuhan University-Baidu Map Beidou Cooperative High-Precision Positioning Technology Joint Laboratory。
文摘In complicated urban environments,Global Navigation Satellite System(GNSS)signals are frequently affected by building reflection or refraction,resulting in Non-Line-of-Sight(NLOS)errors.In severe cases,NLOS errors can cause a ranging error of hundreds of meters,which has a substantial impact on the precision and dependability of GNSS positioning.To address this problem,we propose a reliable NLOS error identification method based on the Light Gradient Boosting Machine(LightGBM),which is driven by multiple features of GNSS signals.The sample data are first labeled using a fisheye camera to classify the signals from visible satellites as Line-of-Sight(LOS)or NLOS signals.We then analyzed the sample data to determine the correlation among multiple features,such as the signal-to-noise ratio,elevation angle,pseudorange consistency,phase consistency,Code Minus Carrier,and Multi-Path combined observations.Finally,we introduce the LightGBM model to establish an effective correlation between signal features and satellite visibility and adopt a multifeature-driven scheme to achieve reliable identification of NLOSs.The test results show that the proposed method is superior to other methods such as Extreme Gradient Boosting(XGBoost),in terms of accuracy and usability.The model demonstrates a potential classification accuracy of approximately 90%with minimal time consumption.Furthermore,the Standard Point Positioning results after excluding NLOSs show the Root Mean Squares are improved by 47.82%,56.68%,and 36.68%in the east,north,and up directions,respectively,and the overall positioning performance is significantly improved.
基金Project supported by the Fundamental Research Funds for the Central Universities,China（No.2572014BB04） the National Natural Science Foundation of China（No.31370710） the Specialized Research Fund for the Doctoral Program of Higher Education of China（No.20110062110002）
文摘Analysis of forest canopy hemisphere images is one of the most important methods for measuring forest canopy structure parameters. In this study, our main focus was on using circular image region segmentation, which is the basis of forest canopy hemispherical photography. The boundary of a forest canopy hemisphere image was analyzed via histogram, rectangle, and Fourier descriptors. The image boundary characteristics were defined and obtained based on the following：（1） an edge model that contains three parts, i.e., step, ramp, and roof;（2） boundary points of discontinuity;（3） an edge that has a linear distribution of scattering points. On this basis, we proposed a segmentation method for the circular region in a forest canopy hemisphere image, fitting the circular boundary and computing the center and radius by the least squares method. The method was unrelated to the parameters of the image acquisition device. Hence, this study lays a foundation for automatically adjusting the parameters of high-performance image acquisition devices used in forest canopy hemispherical photography.
基金supported by the National Natural Science Foundation of China(grant number U20A2091 and No.41771426).
文摘The spatial distribution of power facilities is uneven,thereby making the topology of geographical wiring diagrams(GWDs)based on the actual coordinates unclear.A single-line diagram has the advantage of a clear topology but it lacks spatial locations.A GWD has the advantage of accurate spatial locations but it lacks a clear topology.Visualizing distribution networks for planning requires both features.We proposed a new planning-oriented method for optimizing the visualization of distribution networks.From the global perspective,we proposed an improved force-directed(FD)algorithm by introducing a space restriction strategy and node–edge repulsion strategy to promote the expansion of the distance between distribution facilities within a limited buffer.We then constructed the constrained Delaunay triangulation to identify the compact districts(CDs)and used a genetic algorithm to optimize the parameters for the improved FD algorithm.A novel visualization evaluation indicator was also proposed for quantitatively assessing the visualizations.From a local perspective,the fisheye algorithm was used to optimize the CDs to further improve the visualization of the distribution network.We verified the proposed methods with real-world data.We used limited spatial displacement in exchange for maximum topology clarity to balance the accurate spatial location and topology clarity.
