激光主动成像图像边缘检测算法研究

Edge detection algorithm of intensity image for laser active imaging

针对激光主动成像图像特点,分析了双数复值小波变换的优点和Canny算子存在的不足,提出了一种基于双数复值小波变换和Canny算子相结合的边缘检测算法。首先对激光主动成像图像进行双数复值小波变换得到小波系数序列,再对小波系数进行滤波,最后运用Canny三条检测规则逐层检测图像边缘。通过对实际激光主动图像进行边缘检测实验,结合视觉效果和可匹配性指标,可以看出本文的算法具有图像边缘细节保护好、检测结果可匹配性强等优点。

According to the image characteristics of laser active imaging, the advantages of the dual-tree complex wavelet transform and the disadvantages of Canny operator are analyzed and a new edge detection algorithm based on the combination of dual tree complex wavelet transform （DTCWT） and Canny operator is proposed. Firstly,the laser active image is decomposed into wavelet coefficients with the DTCWT. Secondly, the wavelet coefficients are revised by filter. Finally, image edge is detected according to three Canny detection criterion. Based on edge detection experiments of the actual laser active images, and combined with the visual effect and the matching index, it is observed that the algorithm has the advantages of good image edge detail protection and strong match ability for detection results.