| 摘 要: 针对背景复杂、边界模糊以及芯片相连等特征的芯片Frame图像，提出了一种基于模板匹配的芯片Frame图像分割算法。首先，对整幅芯片Frame图像预分割出多个区域模块；然后，基于区域模块图像采取模板匹配算法匹配单芯片图像；最后，通过合并单芯片的重叠匹配框并记录合并框的坐标信息的方式分割出单芯片图像。实验结果表明：选取合适的模板和阈值，能使该算法的分割准确率达到100%，并且比不基于区域模块匹配的分割算法节省了至少45.76%的分割时间，满足芯片Frame高精度和高速度的分割需求，为芯片图像分割算法的研究提供了一种新思路。 |
| 关键词: 芯片Frame；图像分割；模板匹配；重叠匹配框 |
|
中图分类号: TP391
文献标识码: A
|
|
| A Chip Frame Image Segmentation Algorithm based on Template Matching |
|
YOU Shijun, ZHAO Zilong, ZHANG Jianmin
|
(School of Artificial Intelligence, Jianghan University, Wuhan 430056, China )
ysj919@qq.com; 2416776432@qq.com; zhangjm@jhun.edu.cn
|
| Abstract: Aiming at the chip Frame images with complex background, fuzzy boundary and chip connection, this paper proposes a chip Frame image segmentation algorithm based on template matching. Firstly, the whole chip Frame image is pre-segmented into several region modules. Then, template matching algorithm is used to match the single chip image based on the region module image. Finally, the single chip image is segmented by merging the overlapping matching frames of the single chip and recording the coordinate information of the merging frame.The experimental results show that by selecting appropriate template and threshold, the segmentation accuracy of this algorithm can reach 100%, and it saves at least 45.76% of the segmentation time compared with the segmentation algorithm without region module matching. It meets the requirements of high-precision and high-speed segmentation of the chip Frame. It provides a new method for the research of chip image segmentation algorithm. |
| Keywords: chip frame; image segmentation; template matching; overlapping matching frames |
