Design and implementation of image enhancement based on pyramid architecture

When inspecting radiographies, some detail information cannot be detected due to large range of gray scales and complex tissue structure. The proposed approach use Gauss-Laplacian pyramid architecture to decompose the structure information of the object, then amplify the detail information and resist noise by non-linear processing at different scales. The enhanced detail information is then added back to the original image iteratively. The whole process is similar to wavelet transformation, but the improved image is more consistent to the characteristics of human being´s vision system, thus provides doctors with reliable evidences for diagnosis.