摘要
针对图像中脉冲噪声恢复问题,提出了一种基于像素值分类的脉冲噪声检测与恢复滤波(BPVCF)算法。该算法首先假设图像中像素值为0或255的点是噪声点。以噪声点为中心选取滤波窗口,并对窗口内除中心点外的像素进行分类:存在像素值为(0,10)范围内的点;存在像素值为[10,240]范围内的点;存在像素值为(240,255)范围内的点;像素值全为0和255。然后,根据分类结果,对窗口内像素值为0或255的点作进一步判断,剔除噪声点,保留可疑与非噪声点。最后,根据窗口内剩余点的数目完成对中心点像素值的恢复。试验结果表明,使用该算法处理密度为80%以下的噪声图像时,其主观视觉效果和客观数据评价均优于现有的多种算法,可有效提高机器视觉检测精度。
Aiming at the problem of impulse noise restoration in image,a new algorithm of impulse noise detection and restoration based on pixel value classification filter(BPVCF)is proposed.This algorithm first assumes that the point with pixel value of 0 or 255 in the image is the noise point,and selects the filter window with the noise point as the center,and clasifies the pixels in the window except the center point:there are points in the range of pixel values(0,10);there are points in the range of pixel values[10,240];there are points in the range of pixel values(240,255)and pixel values are all 0 and 255.Then,according to the classification results,the points with pixel values of 0 or 255 in the window are further judged,the noise points are eliminated,suspicious and non-noise points are retained,and the pixel values of the central points are restored by different methods according to the number of remaining points in the window.The experimental results show that the subjective visual effect and objective data evaluation of the proposed algorithm are superior to those of other algorithms when dealing with noisy images with a density of less than 80%.It can effectively improve the detection accuracy of machine vision.
作者
雷继海
LEI Jihai(College of Electrical Engineering,Longdong University,Qingyang 754000,China)
出处
《自动化仪表》
CAS
2020年第12期13-18,共6页
Process Automation Instrumentation
基金
甘肃省工业绿色低碳转型升级基金资助项目(CGLD-2019-049)
甘肃省教育厅高等学校科学研究基金资助项目(2020A-119)。
关键词
图像处理
脉冲噪声
像素值分类
中心像素
滤波窗口
去噪
Image processing
Impulse noise
Pixel value classification
Center pixel
Filter window
Denoising
