针对变光照环境下给皮带撕裂视觉检测过程造成的识别干扰, 提出了一种基于洛伦兹信息测度(Lorentz Information Measure, LIM)分块的改进Ostu分割算法的结构光视觉检测方法。首先将红色线性激光投射到皮带表面, 通过CCD相机捕获高对比度图像; 然后利用LIM方法将ROI(Region of Interesting)区域根据其水平方向光照强度分块, 并逐块通过线性加权Ostu算法进行图像分割, 进而合并出整幅图像的激光条纹; 最后提取激光条纹中心线, 并对其信息特征进行分析比较, 判断皮带是否发生撕裂。通过现场在皮带机上进行试验, 在轻度和重度污染环境下, 892张样本的正确检测率为95.72%, 平均单张样本检测时间为89ms, 检测精度和速度满足实用要求。

In order to solve the interference problem due to variable ambient light in visual detection of conveyor belt rip, a structured light visual detection method based on LIM(Lorentz information measure) partitioning and improved Ostu algorithm is proposed. Firstly, red line laser stripe is projected on the surface of conveyor belt, and high contrast images are obtained by CCD camera. Secondly, region of interest is partitioned horizontally by the LIM data of intensity, and image is segmented by block with linear weighted Ostu algorithm, and laser stripe segmented in these blocks are combine to produce a complete and accurate laser stripe. Finally, center of the laser stripe is extracted, and its feature is analyzed to judge whether there are some rips on the belt surface. According to experimental tests in the scene, the accuracy rate is 95.72%, and the average computational time is 89ms. The requirement of practical use in industry is met.