The invention relates to a multi-vision feature vehicle detection method based on multi-sensor fusion, and belongs to the technical field of unmanned vehicles. The method comprises the following steps: firstly, acquiring image and data information of a forward vehicle through a millimeter-wave radar and a camera, and obtaining a position of the vehicle in the image through a transformation relationship between a millimeter-wave radar coordinate system and an image coordinate system, so as to obtain a region of interest of the vehicle in the image, and reduce the image processing time and interference of a non region of interest of the vehicle on image processing; and then, fusing three attribute properties of the vehicle, namely vehicle bottom shadow, vehicle symmetry and vehicle left and right edges to realize vehicle detection. The multi-vision feature vehicle detection method based on the multi-sensor fusion provided by the invention overcomes the disadvantage that the robustness is low when using a single attribute feature to detect the vehicle.