一种基于HSI的距离联合RGB 3维场边缘探测方法

在彩色图像边缘检测中，欧氏距离方法虽对亮度变化敏感，但对色度和饱和度的差异不太敏感，而矢量角方法又仅能探测色彩差异。为了在RGB3维彩色场中获取精确的边缘信息，提出了一种基于HIS(色度、饱和度和亮度)的距离联合边缘检测方法。该方法在高亮和色差大的区域主要利用矢量角对色彩差异来进行检测，而在低亮和色差小的区域则主要利用欧氏距离的亮度差异来进行边缘检测。实验表明，该方法可有效地解决矢量角存在的问题，其与LUV欧氏距离检测方法相比，则可以避免复杂的色彩空间变换计算，其检测效果也更佳。

A Color Edge Detection Method Using Distance Combination Based on HSI in 3D RGB Color Field

The Euclidean method is sensitive to variations in intensity, but not very sensitive to variations in hue and saturation whereas the vector angle method detects only chromatic differences in color edge detection. The paper presents a distance combination scheme based on HSI (hue, saturation and intensity) so as to capture more accurate edge data in RGB color field. The approach is a metric that the chromaticity difference information in the form of the vector angle is mainly used in edge detection of high intensity and color difference regions, whereas the intensity difference information in the form of Euclidean distance is mainly used in edge detection of areas of low Intensity and color difference. Experiments show that our method effectively solves the problems of unstable angle calculation for vectors with small RGB values, insensitive to intensity differences for vector angle and not very sensitive to chromatic differences for Euclidean in RGB space. Compared to the LUV based measure which has already been used successfully in the LUV space, the method avoids the complexity of the color space transformation and enhances the edge detection over the LUV based results for Visible Man dataset.