基于分块Otsu法的车窗开关缺陷阈值分割方法

Window Switch Defect Threshold Segmentation Method Based on Block Otsu Algorithm

车窗开关的外观与性能有着不可分割的关系,表面缺陷检测是阻止缺陷产品流入市场的重要手段。针对用传统的阈值算法对车窗开关进行分割时容易造成误分现象,提出一种分块Otsu阈值法。对开关组图像去噪后分割开关,然后对开关图像进行分块阈值处理,通过提取特征进行缺陷判断。分块阈值处理是将图像分块,以整幅图像灰度最高值作为基准,对每块图像进行灰度补偿,并进行Otsu阈值法处理。实验结果表明:分块Otsu法能有效分割出车窗开关的缺陷,同时消除了光照不均现象的影响,充分验证该方法的可行性;利用多种常见阈值分割处理进行对比实验,其效果优于其他处理方法,充分验证方法的高效性。

There is an inseparable relationship between the appearance and performance of the window switch.Surface defect detection is an important means to prevent defective products from entering the market.In order to solve the problem that the traditional threshold algorithm is easy to cause misclassification when the window switch is segmented,a block Otsu threshold algorithm is proposed.After denoising the switch group image,the switch is segmented,and then the switch image is processed by block threshold,and the defect is judged by extracting features.The block threshold processing is to divide the image into blocks,to take the highest gray value of the whole image as the benchmark,to make gray compensation for each image,and to process each image by the Otsu threshold algorithm.The experimental results show that the block Otsu algorithm can effectively segment the defects of the window switch,and eliminate the influence of uneven illumination.A variety of common threshold segmentation methods are used for comparative experiments to fully verify the feasibility of the method.