关键词: 主动轮廓模型（ACM）, 图像与视觉计算（IVC）, 灰度增强, 局部灰度特性, 孔洞缺陷检测

Abstract: Considering the incorrect segmentation faults of IVC model segmenting the gray uneven images, this paper proposes an improved algorithm. It removes the gradient information and replaces some factors with constant, which not only decreases the convergence time, but also can segment the gray uneven images accurately. In view of the sunflower seeds’ holes not obvious in the weak light, it enhances its gray linearly and uses the improved IVC model to segment goals, and then combines with local gray characteristics to detect hole defections. The results demonstrate that this algorithm can segment the holes accurately and determine whether its area is hole defects and its average process time is about 30 ms, so it has strong application value.

Key words: Active Contour Model（ACM）, Image and Vision Computing（IVC）, gray enhancement, local gray feature, hole defect detection