Method and device for calculating depth order of monocular image

Abstract

The embodiment of the invention provides a method and device for calculating a depth order of a monocular image. The method includes using a preset over-segmentation algorithm and a preset classifierto sequentially process a monocular image to generate a masking contour map of the monocular image; using a preset convolution kernel to traverse each pixel in the masking contour map to generate a convolution value corresponding to each pixel; in the convolution value corresponding to each pixel, determining the pixels with the convolution value being a preset value as discontinuity points whichare pixels located at the two ends of the missing pixel in the masking contour map; determining a shortest path between the adjacent discontinuity points as a to-be-filled contour between the adjacentdiscontinuity points; along the to-be-filled contour, filling the missing pixel between the adjacent discontinuity points, and generating a filled masking contour map; and calculating the depth orderof the monocular image according to the filled masking contour map. By application of the method and device for calculating the depth order of the monocular image, the depth order of the monocular image can be accurately calculated.

Description

technical field [0001] The present invention relates to the field of computer technology, in particular to a method and device for calculating the depth order of a monocular image. Background technique [0002] In intelligent application technologies such as automatic driving technology and drone technology, the monocular image of the scene can be obtained through the camera, and then the depth order of the scene where the vehicle or drone is located can be calculated based on the monocular image, and finally based on the depth order Carry out scene analysis to realize the automatic driving of vehicles or control the movement of drones. Among them, the calculation of the depth order of the monocular image is mainly divided into two steps: generating the occlusion contour map and calculating the depth order according to the occlusion contour map. The more accurate the occlusion contour map generated for a monocular image, the more accurate the calculated depth order will be....

AI Technical Summary

Problems solved by technology

[0008] However, since the segmentation threshold in the preset over-segmentation algorithm is determined manually, the above method of extracting the occlusion contour of the image cannot be flexibly segmented according to the actual situation of the image, resulting in the occlusion contour in the generated occlusion contour map may be Incomplete, that is, there may be missing pixels in the occlusion contour, which affects the accuracy of the calculated depth order

Images