Abstract: Under the condition that the image brightness environment is relatively poor, the image size is much, and it is relatively coarse. It is needed to compensate the luminance balance and improve the image quality. A multigranulation rough image brightness compensation technology is proposed based on RGB gray decomposition, the collected multigranulation rough image using multi-scale wavelet de-noising method to improve the output image, the noise reduction using RGB decomposition method to get the three components of the gray image, gray image of the three divided components by regular sparse prior overlay method to realize information fusion, through continuous sub block traversal method of image brightness uneven regional white balance compensation, realize the multigranulation rough equalization of image brightness. The simulation results show that the image acquisition conditions under the environment of light at night using the technique has better imaging effect and image output high peak signal to noise ratio, uniform brightness balance, pixel correlation strong visual and imaging performance parameters is superior to the traditional method.