织物组织结构自动识别系统拼接算法研究

针对在织物组织结构识别中，在高放大倍率下，无法在一帧显微图像中获取所观测对象全貌的问题，提出了一种对多帧局部织物图像进行无缝拼接的方法。该方法首先确定织物纱线的周期，再通过动态选取模板进行拼接；并且针对拼接图像较多的情况，采用了以第一帧图像的拼接位置来确定后续图像拼接位置的等距图像拼接方法，从而减小了计算复杂度。该方法已成功应用到织物组织结构自动识别系统中，取得了较快的拼接速度和较好的拼接效果。

On Mosaic Calculating Method of Automatic Identifying System for Fabric Contexture Model

To the problems in the identifying system for fabric contexture model, a method of conducting a seamless mosaic on a number of partial fabric images in a micrograph under high microscopic enlargement times is proposed. On the basis of fixing the fabric yarn cycle, a calculating method is presented to conduct mosaic by selecting templates dynamically. Besides, in response to the large number of images, the equidistant image mosaic method is adopted to simplify the calculation which determines the mosaic positions of the following images according to the first one. This method is successfully applied into the automatic identifying system for the fabric contexture model, and a higher mosaic speed and a better mosaic effect are achieved.