3D reconstruction of indoor scenes is a hot research topic in computer vision. Reconstructing fast, low-cost, and accurate dense 3D maps of indoor scenes have applications in indoor robot positioning, navigation, and semantic mapping. In other studies, the Microsoft Kinect for Windows v2 (Kinect v2) is utilized to complete this task, however, the accuracy and precision of depth information and the accuracy of correspondence between the RGB and depth (RGB-D) images still remain to be improved. In this paper, we propose a post-rectification approach of the depth images to improve the accuracy and precision of depth information. Firstly, we calibrate the Kinect v2 with a planar checkerboard pattern. Secondly, we propose a post-rectification approach of the depth images according to the reflectivity-related depth error. Finally, we conduct tests to evaluate this post-rectification approach from the perspectives of accuracy and precision. In order to validate the effect of our post-rectification approach, we apply it to RGB-D simultaneous localization and mapping (SLAM) in an indoor environment. Experimental results show that once our post-rectification approach is employed, the RGB-D SLAM system can perform a more accurate and better visual effect 3D reconstruction of indoor scenes than other state-of-the-art methods.

Indoor scene reconstruction is important for robot positioning and navigation in scenario reconstruction, especially in constructing a semantic map. In previous research, RGB-D cameras have been utilized to obtain a semantic map. However, because of indoor objects and depth sensors, the accuracy and precision of the depth values could be improved, which is a key factor in reconstructing indoor scenarios. Moreover, there is a relationship between reflectivity and depth accuracy. Therefore, to obtain depth information that is better than that obtained in our previous research, we present a smart post-rectification algorithm based on an artificial neural network (ANN). The algorithm improves the accuracy and precision of depth values by simultaneously considering reflectivity, distances, and different mechanisms of measuring depth. First, we analyze the RGB-D cameras’ characteristics, including the pinhole camera model, lens distortions, and the types of error factors due to the types of RGB-D cameras used. Then, this paper proposes a smart post-rectification algorithm for depth images based on an ANN considering the depth error caused by reflectivity, the distance-related depth error, and different mechanisms for measuring depth. Finally, we perform experiments to evaluate the accuracy and precision of the proposed post-rectification approach by using different types of depth sensors. To evaluate the performance of our proposed algorithm, the proposed approach is applied to RGB-D SLAM, which is tested in different indoor environments. The experimental results show that applying our post-rectification algorithm to indoor scenario reconstruction can result in more accurate and more detailed 3-D reconstruction of objects than other state-of-the-art methods, highlighting the robustness and efficiency of our proposed algorithm.