Solar Cells Inspection has an important position and role in its production. Based on image acquisition and computer vision technology, an automatic inspection method for solar cell surface crack was proposed. Through a series of image pre-processing methods to reduce noise and improve the post-processing capabilities. On this basis, using Gabor wavelets and LGBPHS method to obtain image features, in addition also need to enhance the feature, then using an improved Adaboost algorithm to determine the type of solar cell images whether it has the surface crack. The experimental result shows the method has an efficient and robust effect on automatic inspection of surface crack in solar cell images.