Abstract: In order to eliminate a lot of plaque noises with irregular shape, irregular distribution and unbalanced energy included in video surveillance image of γ radiation environment, a method was proposed to remove the plaque noise by using the correlation information of the local neighboring pixel of the image. Firstly, according to the characteristic of the plaque noise pixel value mutation in γ radiation environment video surveillance image, the plaque noise area was segmented. Then the plaque noise area was repaired by a stepbystep method with the pixel value of the two clean areas as the reference and with the pixel value difference as the correlation degree. In the procedure, a local correlation characteristic between a plaque noise area and neighboring clean area in the frame and another local correlation characteristic between the plaque noise area in the frame and the pixel of the same corridor clear area in the adjacent frame were used. The experiment results show that compared with the traditional image plaque noise removal algorithm, this method has a greater improvement in quantification indicator and visual perception. This noise removal method can effectively remove the plaque noise of γ radiation environment video surveillance image and retain the image detail information.