Monocular image depth recovering method, device thereof and computer equipment

Abstract

The invention discloses a monocular image depth recovering method, a device thereof, computer equipment, a computer-readable storage medium and a computer program. The method comprises the steps of performing characteristic extraction on the monocular image, thereby obtaining the characteristic image of the monocular image; decoupling the characteristic image for obtaining a scene structure graphof the characteristic image; performing gradient sensing processing on the characteristic image and the scene structure graph, thereby obtaining an area-enhanced characteristic image; and performing depth estimation according to the area-enhanced characteristic image, thereby obtaining the depth image of the monocular image.

Description

Technical field [0001] The present invention relates to the field of computer vision, and in particular to a method and device for depth restoration of monocular images, computer equipment, computer-readable storage media, and computer programs. Background technique [0002] The depth recovery of monocular images refers to the depth recovery of monocular images. The depth recovery of monocular images is an important issue in the field of computer vision, and has important applications in many fields, such as three-dimensional reconstruction, robot real-time positioning, and vision Obstacle avoidance and other fields. [0003] However, the depth recovery of monocular images is a complex multi-solution problem due to the uncertainty of the depth scale. Many existing depth estimation schemes often simply couple the scene structure estimation and the depth scale estimation together to solve the problem at the same time, making the problem solving process relatively difficult, often re...

AI Technical Summary

Problems solved by technology

[0003] However, depth restoration from monocular images is a complex multi-solution problem due to the uncertainty of the depth scale.

Many existing depth estimation schemes often simply couple scene structure estimation and depth scale estimation together to solve the problem at the same time, which makes the problem solving process relatively difficult, often requires more data and training time, and has low accuracy.

Images