介绍了光学成像系统灰度衰减模型，提出了一种快速的灰度不均图像校正方法。该方法不需要其他的外部参数，也不需要模型，仅通过对灰度不均的图象自身进行运算来加以校正，它基于通过原图象减去背景图像来消除不均的策略，根所图象特性，直接从原图象中取出属于背景上的点，再通过插值或曲面拟合等方法得到整个背景图象。该方法不但保持了原图象上的目标，而且效果明显，简单快速，并具有很好的通用性，高速公路路面病害检测实验结果验证了该方法的有效性。

The degenerate illumination model for the optics camera system is introduced, and a fast method for gray-level correction is presented. It needn't additional parameters and mathematic model, only performs on the image. It corrects degenerate illumination based on the strategy of original image subtraction from background image. Generally say, the method includes three steps: first, background subset extraction. The image is divided into blocks, then according to the image characteristic, one point is extracted as the background point for each block; second, the whole background building. The whole background image can be got by interpolation from the background subset extracted in previous step; then the last step is correction using subtract the background image from original image This method can not only keep the object of original image, but also be efficacious, fast and simple. It has proved good result from the crack detection of surface of highway.