输送带故障在线检测非均匀光照图像校正方法

Correction method of non-uniform illumination image for on-line fault detection of conveyor belt

提出一种输送带故障在线检测非均匀光照图像快速校正方法.该方法根据线阵相机采集输送带图像的特点,通过曲线拟合统计法估计出输送带图像的行光照度分布曲线,求得图像行中各像素点灰度值的非均匀光照校正系数,校正后的像素值为其原像素值与校正系数的乘积.采用改进的自适应伽马函数,增强图像细节对比度.结果表明:该方法估计的光照分布曲线与Retinex变换法相近,比Retinex变换法等方法实时性强,图像质量好.采用线阵相机在线检测非均匀光照图像快速校正,解决了输送带在线检测中图像光照不均、图像质量差等影响故障检测与识别的问题.

A method to quickly correct the non-uniform illuminated image of conveyor belt was proposed.According to the characteristics of conveyor belt images collected by a linear array camera,such method estimates the line illuminance distribution curve of images through a fitted curve statistical method and works out the non-uniform illumination correction coefficient of grey level for each pixel in image lines.The corrected pixel value is the product of its original pixel value and the correction value.And finally,the modified adaptive gamma function was adopted to improve the detail and contrast of image.The results indicate that the illuminance distribution curve estimated by such method is close to Retinex transformation approach,but has stronger instantaneity,better image quality than methods such as Retinex transformation approach,and solves the problems of non-uniform illumination,poor image quality and bad influence on the failure detection and identification of the on-line fault detection for conveyor belt,and suits to the quick correction of the on-line detection of non-uniform illuminated imagecollected by a linear array camera.