X-ray Image Enhancement Framework Based on Improved Local Adaptive Contrast Field for Complex Workpieces

X-ray images of complex workpieces generally exhibit low contrast using a highly dynamic X-ray imaging system, which reduces the detection sensitivity for minor defects. In previous work, the combination of local variance and variational methods was used for image enhancement, but failed to achieve a good compromise between image enhancement and noise. In this work, an X-ray image enhancement framework based on an improved local adaptive contrast field is proposed to improve the visual quality of X-ray images. The improved contrast field is implemented by constructing a local adaptive gain function, where an improved local variance is used to quantify the fluctuation degree of local information and to reduce the noise in an image. Specifically, the improved local variance is computed with the mean squared error between the real surface and a smooth reference plane determined by Taylor's theorem. Furthermore, an objective function between the improved contrast field and the objective image is solved using a variational approach to obtain a higher-quality image with tiny details highlighted. Experiments with three typical complex workpieces were performed, and results verified the effectiveness of the proposed approach for image enhancement and minor defect detection.