What's it about?

Stereo vision allows capturing of 3D objects or scenes from two images. The core of this technology is the stereo-matching method, which establishes point correspondences between the given images. This paper proposes a new stereo-matching method by locally adaptive image windows.

Why is it important?

The proposed method can determine point correspondences even in low-texture image regions. Furthermore, unknown correspondences by occlusion are estimated using a simple post-processing task. According to the obtained results, the proposed method outperforms similar state-of-the-art methods.

Related papers

• Adaptive matched filter for implicit-target recognition: application in three-dimensional reconstruction


• Facial landmark detection and tracking with dynamically adaptive matched filters