基于像素差异的金属腐蚀区域分割仿真研究

Metal Corrosion Area Segmentation Method Simulation Based on Pixel Difference

研究基于图像的金属腐蚀区域分割问题,提高分割区域的准确性。针对传统基于金属本身与腐蚀颜色的区别,利用像素差异的变化进行腐蚀区域的分割,因当腐蚀颜色与金属的本体颜色相差不是很大,或者在边沿发生像素的混淆时,利用差异边缘像素相减的方法很难完整分割腐蚀区域的问题。为了解决上述问题,提出利用空间像素差异距离聚类的金属腐蚀区域图像分割算法。通过计算差异像素之间的差异距离概率,将差异像素特征进行严格划分,对划分后的结果进行排序计算,详细描述混淆边沿信息,对腐蚀区域进行完整分割。实验证明,改进方法能够对金属腐蚀造成的细微区域完整分割,取得了满意的效果,为金属腐蚀区检测提供了参考。

Research metal corrosion based on image segmentation to improve the accuracy of the segmentation area.Traditional method make use of the difference colours between the metal and corrosion pixels to carry out the image segment.When the color difference in metal and corrosion is not distinct enough,it is difficult to segment the image.In order to solve this problem,the paper proposed an image segmentation method based on the space pixel distance clustering.Through the calculation of the difference distance probability between different pixels,the different pixels were strictly divided.The division of results were ordered,the detailed information of the edges were described,then the corrosion area was completely segmented.Experiments show that the method is capable of metal corrosion segmentation.