基于对比度增强与最大熵的工件表面图像分割算法

Workpiece surface image segmentation based on contrast enhancement and maximum entropy

针对工业相机采集到的工件表面图像缺陷细节不突出、噪声过大等导致最大熵分割法分割不精确的问题,提出一种将多尺度top-hat和形态学开闭重构相结合,并用最大熵法进行分割的工件表面图像分割算法。首先,将工件图像经多尺度top-hat变换得到对比度增强后的图像;其次,将增强后的图像经过形态学开闭重构得到工件图像背景模型;然后将增强后的图像与重构后的图像融合,最后作最大熵分割。实验表明,该算法不仅能准确分割出工件表面的缺陷,并且使细节部分更加突出,同时抑制图像中的大量噪声。

Aiming at the problem that the defect details of the surface image of the workpiece collected by the industrial camera are not prominent and the noise is too large,which leads to the imprecision of the maximum entropy segmentation method,this paper proposes a surface image segmentation algorithm that combines the multi-scale top>-hat with morphological open-close reconstruction and USES the maximum entropy method to segment the workpiece.Firstly,the workpiece image is transformed by multi-scale top-hat transform to obtain the contrast enhanced image.Secondly,the enhanced image is morphologically reconstructed to obtain the background model of the workpiece image.Then the enhanced image is fused with the reconstructed image,and finally the maximum entropy segmentation is made.Experiments show that this algorithm can not only accurately segment the defects on the surface of workpiece,but also make the details more prominent and suppress a lot of noise in the image.