Abstract: An image inpainting algorithm based on improved Laplace operator and cartoon component was proposed in this paper. First, by solving the anisotropic diffusion equation, the cartoon component of the image is obtained. The Gauss low pass filter is coupled with the Laplace operator to generate an improved Laplace operator, the improved Laplace operator combined with the cartoon component, the detection equation of the illumination line is established. Then, dynamic sampling mechanism is used to set up the sample area in order to improve the efficiency of the algorithm. Finally, the normalized root mean square function is established to measure the similarity between the target block and the candidate block, and to complete the image restoration. The experimental results show that: compared with the control group, this algorithm had better visual effect and effectively reduces the blurring effect in inpainting of the complex texture image.