In the process of laser ultrasonic visualization, the external noise and incident wave lead to lower image quality and defect identification capability. In this paper, local statistic filtering is used to reduce the interference of speckle noise, and average gradient and entropy are investigated in order to evaluate denoising results. Two-order differential operator is applied to the enhancement of denoised image, and the image enhancement effects with different differential operators are compared. The penalty function is used to restrain the interference of incident wave on defect scattering wave. The variation trend of peak signal-to-noise ratio and structural similarity are investigated by weighting factor and regularization coefficient, and the method is compared with the method of suppressing incident wave by subtraction of adjacent waves. The results show that the laser ultrasonic visual image quality can be improved by image processing method, and the defect information can be highlighted.