The present invention discloses an artificial compound eye camera three-dimensional image processing method based on geometrical information. The method comprises the following steps of: extraction ofartificial compound eye camera channel images; flat field correction of each channel image; calculation of image space coordinates; calculation of image matching cost; calculation of initial depth; image reconstruction; reconstruction image over-segmentation; correction of the initial depth; and output of reconstruction images and a depth map. The method provided by the invention allow the depthestimation and the reconstruction effect of the artificial compound eye camera to be better, and can be used in the artificial compound eye camera three-dimensional reconstruction field.