Depth map generation method based on user perception optimization

The invention discloses a depth map generation method based on user perception optimization, and the method comprises the following steps: inputting a plane or 2D image, generating a rough depth map through employing a neural network algorithm, marking the region of an object in the image through employing the neural network algorithm or manual operation, and generating a segmentation map, the region of the object being a segmentation block; calculating the features of the image, and calculating an attention score by combining the features of the image; performing energy function design through the rough depth map and the attention score to obtain an optimized depth map; and performing iterative adjustment on the optimized depth map to solve the contradiction of boundary depth difference of the segmentation blocks, and outputting the obtained depth map. According to the invention, a stereo camera is not needed, a common plane picture can be input, a depth image with a better effect can be generated, stereo im