Weak image has characteristics such as low contrast, high noise and poor quality, which affected image observation and application to certain degree. Therefore a weak image enhancement algorithm in wavelet domain was proposed through separating the dimensions of the wavelet coefficients by using multi scale and multi resolution wavelet transform from weak image, histogram equalization of low frequency wavelet coefficients, and edge extraction of wavelet coefficients by using Canny operator. Finally, the image enhancement was realized through reconstructing image by combining each dimension of wavelet coefficients. Three pieces of weak images were selected and compared with the traditional image enhancement algorithm. The simulation results showed that three pieces of weak image through this algorithm showed more details in the subjective evaluation, while the visual feeling smoother and more natural. The information entropy of objective evaluation index of value was also the biggest among several classic image enhancement algorithms with the information entropy values being 4.989 3, 3.741 5, 4.796 1 respectively. The PSNR and CE data showed that the proposed algorithm was moderate with the better overall performance. Therefore, enhancement algorithm aimed at weak image was a relatively good image enhancement algorithm, and it achived the effective image enhancement in the visual effect and image information.