摘要
针对地板砖色差检测的问题,基于统计方法的模式识别理论,提出一种以彩色图像的聚类分析代替人工色差检验的方法。该方法以地板砖彩色图像为研究对象,提取颜色和纹理作为其描述特征,并在此分析基础上,以改进的集合覆盖算法进行聚类分析——该算法用类内最小距离最大的原则选取初始聚类中心,用重心法完成聚类识别。最后用实验结果证明了该方法的可行性和有效性。
The paper provided questions from the different color detection of floor-tile and presented a method that use color image cluster analysis as a substitute of manual detection. The method extracts color and texture as its characteristics description based on color image, selects the initial cluster center the principle of maximizing minimum discrepancy on the basis of SCA algorithm, and completes clustering with Center of Gravity Method. The experimental results show that the method is useful, rapid and satisfactory.
出处
《微型电脑应用》
2009年第11期34-35,5,共2页
Microcomputer Applications
