Abstract: In industrial X-ray testing, owing to the influence of X-ray imaging systems, the environment and other factors often cause the image degradation. By analyzing on image degradation process and aiming at the condition that the detector is linear shift variant system, using of the approximate point spread function algorithm to optimize the X-ray image was used on the basis of inverse filtering signal restoration, and then followed by the filtering for image by KNN algorithm. The experiment results showed that this method was superior to the homomorphic filter in detection effect of X-ray image optimized, and this method also approximately transformed inverse filtering algorithm into division algorithm and greatly reduced the computing time so it was in favor for the application of real-time system.