.png){kind=icon}
.png){kind=icon}
Return
CSIE: Coded strip-patterns image enhancer embedded in structured light-based methods
.png){kind=icon}
PRE
AI
DOI：10.1016/j.optlaseng.2023.107561.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
When a coded strip-patterns image (CSI) is captured in a structured light system (SLs), it often suffers from low visibility at low exposure settings. Besides degrading the visual perception of the CSI, this poor quality also sig-nificantly affects the performance of 3D model reconstruction. Most of the existing image-enhanced methods, however, focus on processing natural images but not CSI. In this paper, we propose a novel and effective CSI en-hancer (CSIE) designed for SLs. More concretely, a bidirectional perceptual consistency (BPC) criterion, including relative grayscale (RG), exposure, and texture level priors, is first introduced to ensure visual consistency before and after enhancement. Then, constrained by BPC, the optimization function estimates solutions of illumination with piecewise smoothness and reflectance with detail preservation. With well-refined solutions, CSIE results can be achieved accordingly and further improve the details performance of 3D model reconstruction. Experi-ments on multiple sets of challenging CSI sequences show that our CSIE outperforms the existing used for natural image-enhanced methods in terms of 2D enhancement, point clouds extraction (at least 17% improvement), and 3D model reconstruction.
Keywords:
Illumination and reflectance estimation
Iterative optimization
Bidirectional perceptual consistency
Coded strip-patterns image enhancer (CSIE)
3D model reconstruction
Journal
IF:
3.7
Papers:
7.1K
Citations:
1.7W