Current Issue Cover
基于小波多尺度分析的图象快速匹配模型

刘海鹰1, 黄胜华1, 彭思龙2, 洪继光3(1.中国科学院自动化研究所,北京 100080;2.中国科学院数学研究所,北京 100080;3.国家专用集成电路设计工程技术研究中心,北京 100080)

摘 要
提出了小波多尺度分析理论在计算机中层视觉二维图象识别中的一种应用,并给出了一个高效匹配模型。利用二维图象的小波多尺度分解来模拟2个最主要的生物视觉通路,并对不同通道分别进行匹配计算,由粗到精地实现了基于扫视和凝视的二维图象的快速识别。与传统的模板匹配法相比,该方法具有速度快、搜索范围大、精度高、抗噪能力强等优点
关键词
Fast Image Matching Model Based on Wavelet Multiscale Analysis

()

Abstract
In this paper, an effective matching model for intermediate level computer vision is proposed based on wavelet multi scale analysis. Two main vision paths are simulated by wavelet multi scale decomposition of images, and scanning and fixing algorithms are designed With complete the fast recognition from coarse to fine scale. Compared with conventional methods, the model is faster, more accurate, and higher noise robust.
Keywords

订阅号|日报