Medical image enhancement method based on visual attention mechanism
Ning Li, Jianyu Zhao, Ping Jiang, Chunmei Li · 2018
A medical image enhancement method based on visual attention mechanism is proposed in view of the characteristics of low contrast and fuzzy medical image. First, this method uses visual attention computing model, analyzes the image significance, and obtains the global saliency map. Second, the global saliency map is divided into a number of subregions, and the average significant value of each subregion is obtained. Third, the weighted statistical coefficient of each area is obtained, the weighted coefficients of each block are multiplied by the statistics of the corresponding region pixels, that is, a histogram that reflects the size of the image information has been formed. Last, the dynamic range of the gray level is adjusted by the visual shielding effect and the principle of histogram equalization, the image enhancement is completed. Theoretical analysis and simulation results show that the algorithm has good enhancement effect on medical images, and can highlight the target area of images and maintain smooth areas.