摘要
传统的小波边缘检测算法在处理边缘定位精度和检测能力方面存在不足,对其进行优化和折衷是提高检测性能的关键,深入研究了小波变换在图像边缘提取中的应用,分析了小波变换时频域分辨率可调和多尺度分析的特性,提出了一种改进的基于相邻尺度小波系数相乘的边缘检测方法和改进的阈值构造方法,通过具体实验数据分析了小波分解尺度和阈值选取与边缘定位精度、边缘漏检误检的关系,以及对抗噪性能的影响。并通过算法仿真,与传统的小波边缘提取方法进行了效果对比。实验表明,针对典型的参考图片取得了较好的边缘检测效果。
A normal wavelet edge detection algorithm is difficult to deal with the gaps in the edge positioning accuracy and detection capability. How to optimize them and make tradeoff is the key to improve the detection performance. The paper studied wavelet transformation employed in image edge detection, and analysed the characteris- tic of adjustable time and frequency region resolution factor and multiple scale character in wavelet decomposition. Then an improved wavelet scale multiplication method and an optimized threshold structure were proposed. The con- crete experiment data under a noise condition explained the change of border fixing position, border missing and bor- der cheat probability related to different decomposition scales and thresholds. Compared with traditional wavelet method this algorithm can get better edge detection result for a representative reference image.
出处
《计算机仿真》
CSCD
北大核心
2009年第10期244-247,共4页
Computer Simulation
关键词
边缘检测
漏检误检
阈值选取
多尺度相乘
Edge detection
Edge miss and false edge
Threshold selection
Scale multiplication
