With the aim to recover the lost information of the apple harvesting robot’s vision caused by the branches and leaves, an automatic image inpainting algorithm was proposed. By calculating the so called blocked factor, the area blocked by the branches and leaves was confirmed at first. And then, a Lattice Boltzmann method based on image inpainting algorithm (LBMII) was used to recover the lost information. The examination results show that the LBMII can recover the apple image effectively, and has higher PNSR than the algorithm based on the curvature-driven diffusions. More over, the parallism of the LBMII can be used to build the high speed parallel image processing system for apple harvesting robot. 