Depth Map Enhancement through Weighted Guided Image Filters in Shape-From-Focus

Abstract

Typically, shape--from.-focus (SFF) approaches do. not take into account any. prior information in order to improve the depth map's accuracy. Estimation of depth maps delivers a key role in the reconstruction of 3D shapes. There are many monocular approaches that use image focus to reshape 3D shapes, and shape from focus is one of them. It uses information about the focus of the optical system to provide a means of measuring 3D information. This study proposed a framework for the enhancement of the depth map by using a weighted combination set of guided filters in shape-from-focus. It has been observed that a different set of weighted combinations of guided image filters are effective in enhancing depth maps in SFF. After evaluation, it is found that the weighted combination of a set of 2 guided image filters provides an enhanced depth map. In comparison to a recent study in which the authors employed a set of 19 filters to enhance the depth map findings. The proposed study gives better outcomes with a faster and less computations-based framework to boost the depth map. In the literature, many guided image filters have been proposed to enhance the depth map individually, but few of them have computational time flaws, and some have unsatisfactory results. A weighted combination of 2 filter sets has been obtained best filter set combination for enhancement of the depth map after evaluation. The optimized weights are obtained using the particle swarm optimization approach, and the subset of best-performing filters is identified through a sequential forward search method. The experimental results have demonstrated that the proposed framework provides considerably improved depth maps, yielding 93% correlation and 4.7 root mean square error to the actual depth map.