In order to solve the overlapping problem in the middle area of images captured by binocular camera in the textile detection system, a mosaicing method for textile image is presented. Firstly, the influence of absolute difference, difference of squares and normalized correlation coefficient on similarity measurement time and accuracy during image registration is investigated. Then, the effect of intergration of different methods-arithmetic average, weighted average and multi-resolution wavelet fusion on the elimination of splice joints is studied. Finally, through analyzing the time complexity, accuracy of matching, and integration effect, the suitable method for fabric image mosaicing is obtained. The results show that the use of gray scale based grille matching coupled with weighted average method is in a good position to complete the fabric image mosaicing.%为解决织物检测系统中双目相机采集图像中间区域互相重叠的问题,提出了用于织物图像的拼接方法.首先研究在图像配准过程中绝对差、差值平方和,以及归一化相关系数对配准时间和精度的影响.其次研究算术平均法、加权平均和多分辨率小波融合方法对消除拼接接缝的效果.最后通过对时间复杂度、匹配度和融合效果的分析,得出适合织物图像的拼接方法.研究发现,采用基于灰度空间上的网格特征匹配和应用加权平均方法进行重叠区域的图像融合可以很好地完成织物图像的拼接.
展开▼
