在平板电脑组装过程中,重要零部件背光板表面往往沾有微型异物,如灰尘,这些异物不易依靠人眼检测,且对最终产品质量有很大影响. 对此,文章提出了一个基于OpenCV与机器视觉的背光板表面异物检测算法. 首先,对待检测零部件图像进行最大类间阈值分割处理与形态学处理,得到包含背光板待检部分的区域;随后利用OpenCV轮廓查找函数cvStartFindContours,定位背光板待检部分的最大外接矩形区域,并联合OpenCV的cvFloodFill函数对背光板轮廓数组内进行置白,生成掩码模板,进行图像按位相与运算处理,从而提取出不规则的待检测区域. 再将提取出的ROI复制一份给RGB图像,且对ROI图像进行开运算处理,对两幅图像进行线性相减,使异物处明显化. 最后,基于图像卷积处理,进一步突出异物,完成异物检测. 实验测试结果表明:与当前图像异物检测算法相比,文章机制具有更好的检测定位效果,准确识别出背光板异物.%Assembled in flat computer process, important parts of the backlight board surface often stained with micro foreign bodies, such as dust, these foreign bodies are not easy to rely on eye detection, and has a great influence on the quality of the final product. Therefore, this paper proposes a backlight plate surface based on foreign recognition mechanism and defect detection of OpenCV. First of all, treat the detected im-age maximum between class threshold segmentation and morphological processing, including the backlight board inspected region. Then, the largest outer OpenCV contour search function of cvStartFindContours po-sitioning to be detected based on the backlight board part of the rectangle region, cvFloodFill function of OpenCV on the backlight plate profile in an array of the white based, generated mask template, image and processing, to extract the irregular tested area. Finally, the extracted ROI copy of a RGB image, and then the ROI image to open operation treatment, two images are linear subtraction, so that the foreign matters at the obvious. Then image convolution based processing, further highlight the foreign body, and in the origi-nal image with a red circle labeling, to display the detection result. The final test performance, the foreign body recognition results show that:compared with the current image recognition algorithm in this paper, for-eign body, mechanism has better recognition effect, accurately identify the backlight board foreign body.
展开▼
