摘要： 针对视觉识别过程中物体材质难以被识别的问题，提出一种基于物体表面光学特征的识别算法。以样本的H, I, S分量为坐标形成HIS颜色空间，通过连接处理，使样本曲面尽量连续。利用边缘检测方法去除被检测图像中的杂质点。求出剩余像素的S分量到该曲面的平均距离，作为判断图像中物体材质和样本材质相似度的标准。实验结果证明该方法简单有效。

关键词: HSI颜色空间, 材质, 边缘检测, 识别

Abstract: Aiming at the problem that object material is difficult to be identified during the vision identification process, this paper proposes an identification algorithm based on the surface optics feature of object. It uses H, I, S——the components of swatch as the coordinate to compose HIS color space, and local continuous curved surface is obtained by connecting. It uses edge detection methods to remove the impure pixels in the detected image. The average distance between the S-component of the remained pixels and the above-mentioned curved surface is ciphered out and regarded as the standard of similar degree for identifying object material and swatch material in the image. Experimental result proves that the method is simple and effective.

Key words: HIS color space, material, edge detection, identification

中图分类号: 

• TN911.73