摘要
针对一类特殊小波变换系数不同尺度下的过零点能提供在不同尺度下信号急速变化点的位置信息,提出了一种基于小波变换的多尺度过零点的双目立体视觉模型的同名点匹配方法,该方法采用由粗到细的匹配策略,利用转化的距离信息先匹配大尺度的小波变换系数的过零点,记录其过零点位置信息,然后应用顺序约束,即物体在左右影像点的投影顺序的一致性,匹配小尺度的小波变换系数的过零点,降低了同名点匹配搜索的复杂性,减少了误匹配,解决了同名点准确匹配难、计算量大等问题。采用该方法对篦冷机内水泥熟料高度进行测量,实验表明该方法能较精确的得出水泥熟料料层的分布状况。
Based on that the multi-scale zero-crossings of wavelet transform, a method is presented to solve correspondence points matching in binocular stereo vision measurement system. This method can obtain the locations of the signal sharp variation points at different scales for a particular class of wavelets. Through studying the physical model of binocular stereo vision measurement system, a coarse-to-fine matching algorithm is applied to solve the problem of correspondence points matching. First, we match the bigger-scale zero-crossings using the information of converted distance, register the location information of zero-crossings, and then match the smaller-scale zero-crossings using sequence restriction, namely the coherence of projection sequence. Thus the complexity of correspondence points matching is decreased and the rate of matching mistake is reduced. The thickness of cement clinker in the cooler is measured using this method. The experiment indicates that the distribution of cement clinker thickness can be calculated accurately.
出处
《光电工程》
EI
CAS
CSCD
北大核心
2007年第7期130-134,共5页
Opto-Electronic Engineering
基金
国家重大技术装备创新研制项目(02-ZZ-11)
关键词
光学测量
小波变换
同名点匹配
立体视觉
多尺度分析
optical measurement
wavelet transform
correspondence points matching
stereo vision
multi-scale analysis
