Depth Map Estimation Model with Efficient Feature Extraction Module

This paper proposes a depth image generation algorithm of stereo images using a deep learning model composed of a CNN. The proposed algorithm consists of a feature extraction unit which extracts the main features of each parallax image and a depth learning unit which learns the parallax information using a 3d cost volume of extracted features. The proposed algorithm uses local region extraction modules for feature extraction and estimates the depth of object region more accurately than existing CNN algorithms.