基于改进Canny的激光主动成像图像边缘检测算法研究  

Laser Active Imaging Image Edge Detection Algorithm Based on Improved Canny Algorithm

作　　者：张腾飞 张合新[1] 孟飞[1] 孙鹏 程展欣 ZHANG Teng-fei ZHANG He-xin MENG Fei SUN Peng CHENG Zhan-xin(The Second Artillery Engineering University, Xi＇an 710025, China)

机构地区：[1]第二炮兵工程大学,西安710025

出　　处：《导航定位与授时》2016年第6期57-62,共6页Navigation Positioning and Timing

摘　　要：针对传统Canny边缘检测算法存在的不足,结合激光主动成像末制导的实际特点,提出了一种改进的Canny边缘检测算法。该算法在Canny边缘检测算法的基础上,采用同态滤波和提升小波变换级联的方法代替传统的高斯滤波器;采用3×3邻域的权值梯度计算方法代替原有的2×2邻域差分运算;采用Otsu自适应阈值计算方法,使算法可根据图像自身特点选择最合适的阈值。实验证明:该算法提升了传统Canny算子的抗干扰能力和鲁棒性,能够较好地保存边缘信息,准确检测出激光主动成像图像的边缘,具有较强的自适应性。By analyzing the shortcoming of the traditional Canny algorithm and combining the characteristics of laser active imaging image,an improved Canny edge detection algorithm was proposed. The algorithm was based on Canny edge detection algorithm,using the homomorphic transformation and lifting wavelet transform to replace the traditional Gaussian filter. The extent of filtered image grades is calculated by using an adaptive method of 3×3 adjacent regions gradient.Finally,Otsu algorithm was used to get appropriate thresholds adaptively based on the features of the image. Experiments show the algorithm improves the anti-interference ability and robustness of the traditional Canny operator,doing well in keeping the detailed edges of laser active imaging image and has strong adaptability.

关 键 词：激光主动成像 末制导 边缘检测 CANNY算子 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]