The invention discloses an autologous joint fluid source mesenchymal stem cell repair cartilage injury image segmentation method based on multi-modal magnetic resonance. The method comprises the following steps of S1, collecting a cartilage injury image, S2, carrying out image visualization processing, S3, carrying out image edge detection, S4, extracting characteristic parameters, S5, extractingseed points, S6, selecting non-seed points, S7, updating the regional gray average value and the pixel standard deviation K of all the seed points in a growth region, and S8, finishing image acquisition. According to the method, 15-20 characteristic parameters are extracted from each edge; wherein the information features comprise edge local information features, edge geometric information features and information features between edge adjacent voxels, the features of different edge lines can be fully embodied, the classification screening precision is improved, and the method realizes intelligent selected image segmentation through self-adaptive seed point extraction and algorithm screening, and improves the image processing speed.