复杂背景下圆形物体分割算法

对如何从复杂的工业图像中准确提取圆形目标进行了研究。阐述了传统分割方法应用于复杂图像中提取圆形目标的局限性，提出了基于双阈值结合数学形态学运算的分割算法。首先分别根据两个阈值进行分割得到两幅二值图像，然后根据目标是圆形以及其大致位置等先验知识分别对两幅图进行腐蚀，开启，闭合等数学形态学运算，将高阈值分割得到的图像中的边界信息叠加到低阈值分割得到的图像中，实现对复杂背景图像中圆形物体的准确分割与提取，解决了复杂背景图像中物体识别率低的问题。实验结果表明本方法对复杂背景的图像预处理是有效的。

Circular object segmentation under complicated background

How to discern circular object accurately from the complex industrial image was studied in this paper. Having illustrated the limitations of the traditional ways in image segmentation, a new method based on double-threshold segmenting in combination with mathematical morphology was put forward. First, it was to obtain two binary images from segmenting image by two thresholds. Second, erosion, opening and closing were done to the two images according to some knowledge of the shape and location of the objects in the image that we have already known. Last, the edge information from the image segmented by a high threshold was put into the image segmented by a low threshold. It is the way to segment and discern circular objects accurately from the complex image and can greatly improve the accuracy rate of discerning. The new method works well in pre-processing with complex image.