Real-Time Dense Reconstruction of Indoor Scene

Abstract

Real-time dense reconstruction of indoor scenes is of great research value for the application and development of service robots, augmented reality, cultural relics conservation and other fields. ORB-SLAM2 method is one of the excellent open source algorithms in visual SLAM system, which is often used in indoor scene reconstruction. However, it is time-consuming and can only build sparse scene map by using ORB features to solve camera pose. In view of the shortcomings of ORB-SLAM2 method, this article proposes an improved ORB-SLAM2 solution, which uses a direct method based on light intensity to solve the camera pose. It can greatly reduce the amount of computation, the speed is significantly improved by about 5 times compared with the ORB feature method. A parallel thread of map reconstruction is added with surfel model, and depth map and RGB map are fused to build the dense map. A Realsense D415 sensor is used as RGB-D cameras to obtain the three-dimensional (3D) point clouds of an indoor environments. After calibration and alignment processing, the sensor is applied in the reconstruction experiment of indoor scene with the improved ORB-SLAM2 method. Results show that the improved ORB-SLAM2 algorithm cause a great improvement in processing speed and reconstructing density of scenes.

Feature Matching Combining Variable Velocity Model with Reverse Optical Flow

“…In the follow-up work, an IMU inertial unit and a magnetometer will be introduced to further optimize the matching between two frames, which can improve the robustness and accuracy of the SLAM system. Additionally, the re-identification algorithms [19][20][21] and trajectory prediction algorithm [22] can be incorporated into the system to improve the recall of loopback detection, which will reduce the cumulative error of the pose, meanwhile, real-time dense reconstruction [23], pedestrian dynamic detection [24], and object detection algorithm [25] can be added to accomplish dynamic target rejection, improve localization accuracy, and accomplish more tasks.…”

Research on map construction based on ORB-SLAM2 algorithm with improved keyframe selection

“…Compared with other visual SLAM algorithm schemes, ORB-SLAM2 algorithm has the features of high localization accuracy, superior real-time performance, clear system framework, and support for multiple camera modes, etc., which has been favored by scholars and become one of the mainstream visual SLAM schemes. Qin [16] improved the visual odometry to optimize the cumulative error by addressing the problems such as camera tracking loss and trajectory drift due to the cumulative error of the RGB-D depth ORB-SLAM2 visual odometer, proposed an adaptive thresholding algorithm to extract the feature points from the image, and the proposed method has better robustness and higher image matching accuracy; Niu [17] addressed the problem that ORB-SLAM2 can only solve camera poses to construct maps by ORB feature extraction, proposed a direct method based on light intensity to solve camera poses and construct dense maps, which greatly improves the processing speed and reconstruction density of the scene; Zhang [18] proposed an improved ORB-SLAM2 dense map optimization algorithm in order to quickly obtain accurate indoor 3D dense maps at a lower cost, the constructed maps can be directly used for navigation with higher accuracy, shorter tracking time and smaller memory; Luo [19] proposed an improved ORB-SLAM2 algorithm based on information entropy and sharpening processing for the problem that the SLAM system cannot locate and build a map in the case of motion scene and image texture information is not rich, the relative trajectory error is reduced by 36.1% compared to the original system, and the absolute trajectory error is reduced by 45.1%.…”