Retrieving point cloud models of target objects in a scene from photographed images
- PDF / 7,912,473 Bytes
- 18 Pages / 439.642 x 666.49 pts Page_size
- 76 Downloads / 198 Views
Retrieving point cloud models of target objects in a scene from photographed images Nan Luo1 · Ying Xu1 · Quan Wang1 · Bo Wan1 Received: 20 August 2019 / Revised: 5 August 2020 / Accepted: 15 September 2020 / © Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract Image-based reconstruction is devoted to recovering the 3D point cloud models of target objects from scene images photographed at different viewpoints, and the existing methods often produce a large number of redundant background points, which causes inconvenience to 3D modeling or other related applications. To solve this issue, this work proposes an improved framework that combines image segmentation in the point cloud retrieving procedure, so as it only reconstructs the objects of interest in a scene. This framework provides two options for foreground object segmentation, and users can determine the appropriate method to obtain accurate segmentation for different scenes. Then, the feature matches are extracted from the segmented images, and the point cloud model is recovered via two phases of dense diffusion, feature diffusion and patch diffusion. In the diffusion stage, we introduce a new normalized metric that deals with both the illumination change and low texture case to enhance the robustness of the reconstruction. The experimental results show that proposed framework can effectively avoid reconstructing the irrelevant background data while outputting more even and detailed point cloud models. Keywords Point cloud retrieving · Image sequences · Foreground segmentation · Dense diffusion · Illumination change
1 Introduction Reconstructing point cloud models from image sequences is an important research issue in computer graphics, which provides data support for many practical applications, such as medical, industrial measurement, cultural heritage protection, 3D movie entertainment, and virtual reality. Comparing to the direct 3D scanning devices [12, 35], image-based retrieving method restores the 3D point coordinates of object surface from captured images and has the advantages of flexible, low-cost and strong practicability. It can be applied to various This paper was supported by the National Natural Science Foundation of China(Grant No.61802294), China Postdoctoral Science Foundation(Grant No.2018M633472). Bo Wan
[email protected] 1
No.2 South Taibai Rd, Xi’an, China
Multimedia Tools and Applications
indoor and outdoor environments to reconstruct the realistic object models, showing great research saliency and application potentials. Image-based 3D reconstruction is an interdisciplinary subject involving image processing, stereoscopic vision, and computer graphics. In order to obtain dense point cloud model, the appropriate reconstruction scheme is seed-and-expand which contains three steps: image preprocessing, feature match diffusion, and stereo mapping [33]. Since the captured scene images inevitably include cluttered backgrounds besides the target object, resulting numerous unrelated background points in the
Data Loading...
