摘要
本课题提出了一种适合高速宽幅造纸机纸病在线实时检测的复合算法,该算法的基本思想是:首先通过快速傅里叶变换(fast Fourier transform,FFT)将时域图像转换到频域,以便使用频域图像与高斯(Gaussian)高通滤波器做乘积运算进行滤波;再通过快速傅里叶逆变换(inverse fast Fourier transform,IFFT)将频域中的图像转换到时域,以便下一步使用运算速度较快的分水岭算法对滤波后的图像进行快速缺陷分割,实现纸病在线实时检测。利用采集到的2000余张纸病图片进行纸病检测实验,结果表明,本课题提出的复合算法具有快速、高效、适用性强、分割效果好等优点,可满足纸病在线检测对图像处理算法实时性和准确性两方面的要求。
This paper presented a composite algorithm appropriate for online real-time detection of paper defects in high-speed and widewidth paper machines.The algorithm employed the following basic idea:First,the time-domain image was transformed into the frequency domain using fast Fourier transform(FFT),enabling the product of frequency domain image and Gaussian high-pass filter to be utilized to filter.Next,the image in the frequency domain was converted back to the time domain via inverse fast Fourier transform(IFFT).The third step involved utilizing the fast watershed algorithm to quickly segment the filtered image and achieved online real-time detection of paper defects.Over 2000 pictures of paper defects were collected.The results showed that the proposed composite algorithm possessed several advantages,including high speed,high efficiency,strong applicability,and good segmentation effect.Ultimately,it met the demands of realtime capability and accuracy for online paper defect detection.
作者
汤伟
张龙
王锦韫
方嘉楠
TANG Wei;ZHANG Long;WANG Jinyun;FANG Jianan(College of Electrical and Control Engineering,Shaanxi University of Science&Technology,Xi’an,Shaanxi Province,710021)
出处
《中国造纸》
CAS
北大核心
2023年第7期153-160,共8页
China Pulp & Paper
基金
西安市科技计划项目(2020KJRC0146)
国家自然科学基金计划项目(62073206)。
