Abstract: To overcome the problems of X-ray image's vertical stripes caused by uneven response of X-ray linear detector's pixels and the noise interference produced by detector and outside, a new calibration and filtering model was proposed. It solved the problem that traditional calibration result was very volatile with ignoring noise. By combining X-ray's features and X-ray linear array detector's imaging principle, the output characteristic and noise were analyzed when the pixel responded unevenly. Besides, the calibration and filtering model was built. Improved-two-point calibration algorithm was used to calibrate the image and semi implicit difference method based on partial differential equation was used to smooth the noise. Experiment results show that the unevenly response of X-ray linear array detector with a resolution of 19 216 was eliminated, and the influence of noise was suppressed. In the case of 16 bit per pixel,the mean square error was below five gray level, it improves the quality of X-ray image detection.