受火炮内膛结构、材料与光照等影响，火炮内膛全景图像存在光照不均现象，影响图像的分割质量。采用背景去除法消除内膛图像照度不均，同时在基于小波变换的同态滤波法的基础上，根据贝叶斯萎缩阈值法对其进行了改进，对于经过小波变换后内膛图像的高频子带图像进行加权修正，而对其低频部分进行线性调整。针对校正结果，依据像素信息熵，提出了不同算法对无标准图像改善效果的评价标准。火炮内膛全景图像实验表明，改进算法在保持了原算法大幅度改善图像品质的优点的基础上，减少了边缘信息的丢失。

Influenced by gun bore structure and illumination, the gun bore panoramic image has the problem of uneven illumination which will influence the quality of image segment. The de-background is adopted to eliminate the uneven illumination in gun bore images, and the homomorphic filtering based on wavelet is ameliorated through the bayesian shrinkage method. The high-frequency subband images of gun bore panoramic image are revised by weighting, and the low-frequency subband images are modulated linearly. The evaluation criterion for corrections of nonstandard images by different algorithms is proposed. Experimental result of the gun bore panoramic image shows that the modified algorithm can decrease the expense of marginal information, on the basis of improving the image quality primely.