Abstract: In the case of lowlight conditions, the quality of the output image is not satisfactory provided by complementary metal oxide semiconductor (CMOS) imaging device.An image enhancement algorithm based on Kmeans clustering was proposed according to the characteristics of low illumination image, and the analysiscomparison of various algorithms.The clustering center k can be automatically determined by the histogram feature of original image,and the sub images can be enhanced by histogram equalization according to the information content of each sub image after pixel sets are divided into several nonoverlapping subsets by clustering.The experiment was carried out in CMOS imaging device using this method,the results shows that this algorithm can enhance the contrast of the image to 17 times of the original image, and the average gradient can be increased to 4 times under the condition of preserving the details of the image (information entropy) to about 98.6%.