Learning Dynamic Guidance for Depth Image Enhancement




To better modeling the dependency between intensity and depth map, we proposed a weighted analysis representation model for guided depth reconstruction. An intensity weighting term and an analysis representation regularization term are combined to model complex relationship between depth image and RGB image. We utilized a task driven training strategy to learn stage-wise parameters for specific task, the proposed model is able to generate high quality depth restoration results in a few stages. Compared with other state-of-the-art methods on both the guided depth upsampling and restoration problems, the proposed model achieved better results with less RMSE value and more pleasant visual quality.

ref :

Gu, Shuhang, Wangmeng Zuo, Shi Guo, Yunjin Chen, Chongyu Chen, and Lei Zhang. “Learning dynamic guidance for depth image enhancement.” Analysis 10, no. y2 (2017): 2.

Some image enhancement methods

MATLAB code for image enhancement algorithms and information theory analysis


There are no reviews yet.

Be the first to review “Learning Dynamic Guidance for Depth Image Enhancement”

Your email address will not be published. Required fields are marked *