Abstract: According to the traditional pixel-level image fusion method fragmented issue of linkage between pixels proposed，the image fusion method is put forward，which is based on the regional detection.First of all，through the pixels in its neighborhood of the variance，the way is used to distinguish whether it is clear point or ambiguous point.Secondly，to have a clear area by the current pixel as the center of the floating window，the corresponding window region，a clear statistical points，eliminating the center of fewer pixels clearly points to the same way local window region，a clear statistical points，points more clearly the center of pixels set to a clear point.Finally，according to the results of detection，regional images is integrated.Experimental results show that the proposed method is effective.

Key words: regional detection, image fusion, clear pixels, variance, focus

摘要： 针对传统像素级图像融合方法割裂像素间联系的问题，提出一种基于区域检测的图像融合方法。首先，通过像素点在其邻域内的方差来区分是清晰点还是模糊点，其次，以当前像素为中心的浮动窗口统计相应窗口区域的清晰点数，消除清晰点数少的中心像素，以同样的方式统计局部窗口区域的清晰点数，清晰点数多的中心像素点设置成清晰点，由此得到清晰区域，最后，通过检测出的区域结果进行图像融合。实验结果表明提出的方法是有效的。

关键词: 区域检测, 图像融合, 清晰像素, 方差, 焦点