"双目视觉下的SLAM三维场景建图及物体识别研究-重庆大学硕士学位论文"

The research on 3D scene reconstruction of SLAM and object recognition under stereo vision is a significant topic in the field of mechatronics engineering. This thesis, written by Zhou Quan and supervised by Associate Professor Wang Jian from Chongqing University, provides a comprehensive investigation into the application of stereo vision in simultaneous localization and mapping (SLAM) for building 3D scenes and recognizing objects. The introduction of the thesis outlines the background and significance of the research topic, emphasizing the limitations of traditional monocular vision systems in mapping and recognition tasks. The use of stereo vision, which involves capturing images with two cameras to create depth perception, offers more accurate and detailed information for scene reconstruction and object detection. The methodology section details the process of SLAM algorithm implementation for 3D scene reconstruction, including feature extraction, matching, and triangulation. The research also explores the integration of object recognition techniques using machine learning algorithms to identify and classify objects within the reconstructed scenes. The results and discussion section presents the findings of the study, including the accuracy and efficiency of the 3D scene reconstruction and object recognition techniques. The experimental results demonstrate the feasibility and effectiveness of using stereo vision in SLAM for building detailed and precise 3D maps while accurately identifying objects within the environment. Overall, this thesis contributes to the advancement of mechatronics engineering by showcasing the capabilities of stereo vision in enhancing the accuracy and reliability of SLAM-based 3D scene reconstruction and object recognition. The research opens up new possibilities for the development of robust and intelligent systems for various applications in autonomous robotics, augmented reality, and smart manufacturing.