AN OPTIMIZATION METHOD OF COMPRESSIVE COMPUTATIONAL GHOST IMAGING BASED ON MULTI-SCALE DECOMPOSITION AND FUSION

For acquiring the higher recognition ghost image, we proposed a optimization method to improve the edge and texture details of the ghost image based on the multi-scale decomposition and directional filtering. Firstly, taking the nature of the filters can be divided, a group of speckle patterns is divided into three directions by a filter wheel and three encoded speckle patterns are generated. Secondly, measure and record the total light intensity of speckle pat-terns in different directions that passes through the object, and reconstruct the ghost in three directions respectively by compressive computational ghost imaging. Finally, fuse the ghost image in different directions based on inverse NSCT to improving the edge and texture details of ghost image. A set of simulation and experimental results show that the peak signal-to-noise ratio, average grads and space frequency have been further promoted.