| The high-energy photons in the gamma radiation scene will interfere with the normal imaging of CMOS image sensor,it will cause visual degradation problems such as patch noise and global distortion in the scene images,which seriously affects the observation effect of gamma radiation scene images.Therefore,the enhancement method of gamma radiation-contaminated images based on spatiotemporal correlation is proposed.It is mainly considered from two aspects to achieve a clear visualization of the gamma radiation scene: speckle noise removal and global distortion correction.The main research contents included are as follows.(1)Analysis of the mechanism and characteristics of visual degradation of gamma radiation scene images.The degradation principle and visual characterization of gamma radiation scene images are analyzed from the mechanism of gamma photon interaction with CMOS image sensor.Firstly,the process of Compton scattering when high-energy photons penetrate CMOS image sensor in gamma radiation scene is analyzed.Based on that,the image characterization of speckle noise in scene image is analyzed in combination with image processing methods.Secondly,the physical and visual properties of the global distortion of the image in gamma radiation scene are analyzed by learning the interaction processes between Compton electrons and matter.It lays the foundation for the design and validation of the enhancement method of gamma radiation scene images.(2)Research on speckle noise removal method based on spatiotemporal correlation.Aim at the serious interference for the scene clear visualization produced by speckle noise in gamma radiation scene image.A speckle noise removal method based on spatiotemporal correlation is proposed.Firstly,an image illumination normalization method is used to eliminate the effect of subtle illumination differences in sequence images on noise detection.Secondly,noise detection is achieved by outlier analysis of temporal pixel sequences.Finally,the significant noise in the gamma radiation scene images is removed by finding effective pixels from the temporal pixels,and weak texture noise at the edge of the speckle noise is eliminated by combining spatial structure denoising method.(3)Research on global distortion correction method based on remapping.To eliminate the effect of global distortion in gamma radiation scene images on scene discrimination,a global distortion correction method based on remapping is constructed.Firstly,to globally improve the image contrast,a contrast limited histogram equalization method is used to redistribute the image brightness.Secondly,in order to realize the correction of image color,the color component remapping method is proposed to eliminate the unbalanced dynamic response of each color channel.Finally,according to the visual characteristics of human eyes,segmented nonlinear remapping is constructed to fit the mapping to the light and dark regions of the image.To verify the effectiveness of this topic in the enhancement of gamma radiation scene images.The enhancement method is used for performance verification experiments in real gamma radiation scene images,and also for comparison experiments with several image enhancement algorithms with excellent performance.The comparative experiments on visual effects and quantitative index evaluation demonstrate the scientific design of the method and its effectiveness in image enhancement of real gamma radiation scenes. |
