Aiming at the shortcomings of traditional apple defect detection methods such as high labor intensity, low productivity and high false positive rate,the apple defect detection algorithm is proposed based on visible light-infrared light image fusion. The algorithm uses different fusion methods for high and low frequency wavelet coefficients of visible and infrared images to obtain more prominent feature images. The results of simulations show that the recognition rate can reach 96% in the detection of apple fruit with scratches, bumps, fruit stems/flower buds and intact fruits, in addition, the recognition accuracy rate for scratches can reach 92%, and in the detection of fruit stems, flower buds and intact fruit, the accuracy rate can reach 100%. It fully meets the needs of apple's online detection and classification.