The invention discloses an image fusion method for apple quality detection, mainly to solve the problem that the traditional fruit quality detection is not high in effectiveness. The method comprises steps: 1) quaternion wavelet transform is used for a visible light image and an infrared image respectively to obtain low-frequency sub bands and high-frequency sub bands; 2) a weighted average method based on a local average gradient and a local standard deviation is used for realizing fusion on the low-frequency sub bands; 3) a fusion rule for taking a large absolute value based on steerable filtering is adopted for the high-frequency sub bands to realize fusion; and 4) finally, high-frequency coefficients and low-frequency coefficients after fusion use inverse quaternion wavelet transform to obtain an image after fusion. The detail texture of epidermis in the visible light image and subsurface damage information detected in the infrared image can be well kept, and thus, the method can improve the apple quality detection accuracy.