Denoising Of Real-time X-ray Imaging Technology Research

In the industrial real-time X-ray detection,enhancing the image contrast,removing X-ray image noise are the main research directions of improving penetration sensitivity. In this dissertation, study the system of industrial X-ray imaging based on an image intensifier, make focus on the noise of real-time X-ray imaging, and research the methods of noises suppression.The thesis analyzed the principle of the radiographic imaging and the constitute of imaging system, then analysis the factors of affecting image quality in the imaging process, the divided the image noise into random noise and pixel channel response non-uniformity noise. For the elimination of random noise, the image focuses on internal use of templates smoothing. In order to removing non-uniformity noise, the thesis builds a pixel channel response non-uniformity correction, then correction the non-uniformity by least squares to modify the uniformity respond. According to the sequences images, researching the average of sequences still images and the video signal recursive filtering and time-series images of three-dimensional model of noise study, then bring forward the method of pixel offset of each image relative to reference image are calculated and offset all the images to overlap the identical parts. Through motion estimation, the frame image sequence template matching, and the movement of objects frame image sequences for effective removing noise reduction.Finally, according to the model of pixel channel response non-uniformity correction, the air-screen image and weld image are optimization corrected. Based on the method of offset and average denoising, it is validated by experimental results that penetration sensitivity in real-time weld radioscopy.