基于柔性形态学的抗噪彩色图像边缘检测

Noise-resistance in color image edge detection based on flexible morphology

针对由RGB颜色空间各分量高相关性引起的利用分量边缘检测不能有效滤除噪声以及边缘模糊较粗的问题,选取彩色图像处理复杂性较低并且符合人眼色彩感受的HSI颜色空间,提出一种将柔性形态学用于彩色图像边缘检测的方法.利用小尺度算子改进彩色柔性形态学算子的抗噪性,即先对H、S、I 3个分量使用改进的柔性形态学算子检测边缘,然后将3个分量边缘加权得到融合的彩色图像边缘.结果表明,此方法能够在充分保留彩色图像边缘信息的基础上,有效地抑制噪声,得到连续、完整、清晰的边缘.

In view of the fact that a high RGB color space correlation between the components used in edge detection can effectively filter out noise and blurred edge.A color image processing of lower complexity was chosen that could match the human eye color feeling for HSI color space,and a flexible morphology was put forward for color image edge detection.A small-scale operator was used to improve the color noise resistance of a flexible morphological operator,the first three components of H,S,I were used to improve the soft morphological operator edge detection,and then the fused edge was weighted by the three components of the color image edge.The experiment results showed that this method could fully keep the edge information of a color image,thus being able to effectively suppress noise and obtain a continuous complete clear edge.