基于Haar小波和形状模板的图像快速匹配算法  

Fast Image Matching Algorithm Based on Wavelet and Shape Template

作　　者：白冰峰[1] 温秀兰[1] 张中辉[1] 

机构地区：[1]南京工程学院自动化学院,南京211167

出　　处：《组合机床与自动化加工技术》2017年第2期37-40,共4页Modular Machine Tool & Automatic Manufacturing Technique

基　　金：国家自然科学基金项目(51675259);江苏省"333人才工程"项目资助;江苏省普通高校专业学位研究生创新计划项目(SJLx16_066)

摘　　要：针对传统的模板匹配方法在低对比度的图像中匹配不到目标物体和匹配计算量大等问题,提出一种基于离散Haar小波变换和形状模板的图像快速匹配算法。首先,使用Haar小波对图像进行分解操作,将分解的低频信号重构得到压缩图像,以减少匹配计算量;其次,使用形状模板对压缩图像进行匹配,设计出Haar小波图像压缩和形状模板匹配相结合的算法及程序。最后,经大量实验结果证实:所提出的算法不仅匹配速度快,精度高,而且匹配结果不受遮蔽、混乱和非线性光照变化等情况的影响,适宜于在复杂环境中进行图像提取、工件识别等。According to these problems that the target object is not matched in low-contrast images and large matching calculation is needed in the method of traditional template matching,a fast image matching algorithm based on discrete Haar wavelet transform and shape template is proposed. Firstly,the image is decomposed by Haar wavelet,then the compressed image is reconstructed through the lowfrequency signal of decomposition to reduce computation cost of image matching and the compressed image is matched by shape template,the algorithm and program of Haar wavelet image compression and shape template match are designed. Finally,a large number of experimental results showthat the proposed matching algorithm not only has fast speed and high accuracy,but also the matching results are not influenced by shelter,chaos,the changes of nonlinear illumination and so on. And it is applied to image matting and workpiece recognition in the complex environment.

关 键 词：小波变换 形状模板 图像匹配 

分 类 号：TH161[机械工程—机械制造及自动化] TG506[金属学及工艺—金属切削加工及机床]