基于视觉特性和K-L变换的图像边缘检测  

Edge Detection Based on Visual Properties and K-L Transformation

作　　者：李峰[1] 冯新喜[2] 侯志强[2] 

机构地区：[1]空军工程大学电讯工程学院,西安710077 [2]空军工程大学

出　　处：《微计算机信息》2008年第21期306-308,共3页Control & Automation

摘　　要：文章提出了一种基于人眼视觉特性和K-L变换的彩色图像边缘检测方法。首先将原始图像应用人眼彩色传递模型进行频率域滤波处理,使得图像更加平滑,并减小噪声的干扰。而后根据滤波结果重建图像,用K-L变换处理图像的三个特征量,对主要彩色分量用Canny算子检测边缘。由于人眼的视觉特性近似频域低通滤波器,对使用人眼彩色传递模型预处理后的图像检测边缘能够使得检测结果有较好的连续性和抗噪声性。仿真结果证明该方法能够有效地检测彩色图像边缘。A color edge detection approach based on human visual properties and K-L transformation is presented in this paper. The approach firstly filters the image in frequency field, So, the image is smoother, and the disturbing affects of yawp are Weaken, Then the three characters of the rebuild image are transformed with K-L transformation, The main character is kept for edge detection by using Canny operator. Because the function of human visual properties is like low-passed filter in frequency field, so the results of color edge detection have good continuity and are not sensitive to yawp. Simulation results show that the approach is efficient in color edge detection.

关 键 词：彩色图像分割 边缘检测 CANNY算子 K-L变换 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]