Monocular Depth Estimation Methcd Based on Local Regional Reinforcement
A monocular depth estimation method based on local regional reinforcement was proposed to ad-dress the issues of object boundary distortion and loss of local detail information caused by complex texture and geometry structare in depth estimation scene.First,the depth estimation model based on the convolu-tional neural network was used to obtain the low-resolution image.Then,the saliency object detection model was introduced to obtain the high-resolution saliency map which supervised the generation of depth map.Finally,the salient map and depth map were fused to improve the overall depth estimation accuracy of the image.Experimental results on public datasets show that the proposed method can significantly im-prove the precision of monocular depth estimation.
monocular depth estimationlocal regional reinforcementconvolutional neural network deep learning
国家科技期刊平台
NSTL
万方数据
维普
