改进Canny算法的实时图像边缘检测
胡晓莉，丁 丽，江 杰
（内蒙古科技大学 信息工程学院，内蒙古 包头014010）
关键字：小波变换；基于阈值的极值中值滤波；边缘检测；FPGA
中图分类号：TP391.4 文献标识码：A
摘 要：传统的Canny边缘检测中高斯滤波参数难以设定，且局部噪声干扰难以消除，容易检测出伪边缘。针对这一问题，提出了一种改进Canny算法的实时图像边缘检测方法。该方法应用小波变换去除噪声，利用基于阈值的极值中值滤波代替Canny算子原有的高斯滤波，来提高图像的边缘检测效果。同时，为了快速准确的得到实时图像的边缘检测数据，结合FPGA并行处理能力强、数据吞吐量大的特点，通过在FPGA上构建各模块进行图像处理。实验结果表明，应用该方法可以有效实时地检测出图像边缘。
Real-time image edge detection of theimproved Canny algorithm
HU Xiao-li , DING Li , JIANG Jie
(Information Engineering School , Inner Mongolia University of Science and Technology , Baotou 014010 , China)
Abstract: Gauss filter parameters are difficult to be set in thetraditional Canny edge detection, and it is easy to detect false edge but difficultto eliminate the local noise. Aiming at this problem, an real-time image edgedetection was presented based on the improved Canny operator .The wavelettransform was used to remove noise , and then , extremum median filtering basedon threshold was used instead of the original Canny operator Gauss filter toimprove the effect of image edge detection . At the same time , combined with FPGA’scharacteristics of strong parallel processing and great data throughputcapabilities, the module of image processing in FPGA was constructed in orderto get fast and accurate real-time image of the edge of the image data . Theexperimental results show that the image edge can be effectively detected by usingthis method .
Key words: wavelet transform ; extremum median filtering based onthreshold ；edge detection ; FPGA