基于差分图像的滴塑模具残留物检测算法

Plastic drop mold residual detection algorithm based on difference of image

摘要:

模具是工业产品的成型工具,在滴塑加工过程中,若模具存在异物,它会导致下一次成型时成品出现孔洞,严重降低产品质量。针对该问题,提出了一种基于差分图像的残留检测算法,首先采用基于灰度模板匹配算法定位检测区域,再将标准图像和当前图像作差分,并进行阈值分割和形态学处理获得二值化图像,最后对图像进行连通区域分析,得到残留物在图像中的位置和大小；针对环境光照变化引起的误检现象,采用了一种全局光照补偿算法降低其影响。测试结果表明,该算法效率高并具有良好的稳定性,适用于模具的残留物检测。

Abstract:

Mold is forming tool for industry products. For plastic drop process, the mold residue will cause the worse quality of the product. As for this problem, it proposes a residual detection algorithm based on difference image. First, template matching algorithm based on gray image is used to locate the detection area. Then it calculates the difference of standard image and current image. Binary image is obtained by thresholding and making morphology processing on the difference image. Finally, it calculates the position and size of residuals by connected component analysis. As for mistake detection problem caused by illumination changing , a global illumination compensation algorithm is used to reduce its impact.The test result shows it has high efficiency and stability and is suitable for mold residue detection.