The invention relates to the technical field of navigation systems, in particular to a blind autonomous navigation method based on the stereoscopic vision and the information fusion. The output part of a left image and right image is connected with an image preprocessing part, the output of the image preprocessing part is connected with an image segmentation part, the image segmentation part is connected with a characteristic extraction part, the output of the characteristic extraction part is connected with a stereographic matching part, the output of the stereographic matching part is connected with a parallax chart obtaining part, the output of the parallax chart obtaining part is connected with a depth image obtaining part and a target detection and marking part, the output of the depth image obtaining part is connected with the target detection and marking part, the output of the target detection and marking part is connected with a target size, distance, direction and calculation participation part, the output of the target size, distance, direction and calculation participation part is connected with a multi-sensor information fusion part, the output of the multi-sensor information fusion part is connected with a voice output part of information, such as obstacle position, size, distance, direction and the like, and the outputs of an ultrasonic wave ranging part, an electronic compass direction finding part and a GPS positioning part are connected with the multi-sensor information fusion part. In the invention, the algorithm has great use value for improving the autonomy and the independent walking ability of the blind.