摘要： 在目标大小已知、视点相同的情况下，通过对输入图像中各个与目标大小相同的子窗口进行局部分割，并确定每一子窗口局部分割结果与目标模板间的公共区域，分别提取两者在极坐标下的特征，根据两者特征的相似性，在一定的相似性门限下，实现对子窗口的识别。实验证明，该方法在背景复杂、目标分割质量较差的情况下，仍能取得较好的识别效果。

关键词: 图像识别, 特征提取, 图像分割

Abstract: This paper presents an objects recognition method under given object size and same view point situation, that locally segments the same size window with the template by search the target image, then segments the template, and gets the intersection between template and the searched window. By matching the common pixels between the searched windows and the template via searching the whole image, it maximizes the correlation coefficient or select several maximums, best matching patches gotten.

Key words: image recognition, feature detection, image segmentation

中图分类号: 

• TP391.41