|Thesis Advisor||谭民 ; 喻俊志|
|Place of Conferral||中国科学院自动化研究所|
|Keyword||立体成像 多目视觉 相机标定 图案 偏振|
When acquiring 3D surface information under intracavity circumstances, imaging systems face great challenges because of the small size, texture less, and large flexibility of the environment. Multi-camera polarization vision refers to multi-camera systems with polarizers, which is a hopeful hardware solution to accomplish intracavity 3D imaging based on passive method without feature extraction. To capture the polarization information in images, the system should contain at least 4 cameras, which makes common FOV of cameras narrow and asynchronous calibration of polarizers and cameras. This thesis focuses on camera calibrations based on 2 novel designed patterns and synchronous calibration of the system. Relative topics are all also discussed. The main technical contributions are summarized as follows.
Firstly, a novel pattern design framework and pattern generating algorithms are proposed to solve the incomplete sample problem for calibration pattern. In proposed design procedure, design targets are decomposed and assigned to be realized in three different scales. Based on the design framework, two calibration patterns are generated by the integrated strategy of wave function collapse algorithms and iterative strategy.
Secondly, image processing algorithms are designed for proposed calibration patterns, which contains feature detection, sorting and identification. Feature detection is implemented based on prototype filtering, and both coordinates and states of features are acquired. Feature sorting is based on checkerboard-growing algorithms. Then, feature identification is realized by a reverse procedure of pattern design. Finally, image processing algorithms are integrated into stereo calibration. Experiments show that the utilization of incomplete samples in calibration could improve the accuracy of results.
Thirdly, a synchronous calibration method for multi-camera polarization vision system is proposed. The calibrations of polarizers are conducted based on the multi-camera calibration results, avoiding the possible interactions of calibrations cameras and polarizers. The iterative optimizer is regarded as a controller for a first-order stabilization problem, then disturbance rejection control law is applied in the iterative optimizer, and effectiveness are tested by a verifing experiment.
|高彰. 面向腔内被动式立体成像的多目偏振视觉系统标定研究[D]. 中国科学院自动化研究所. 中国科学院大学,2021.|
|Files in This Item:|
|Recommend this item|
|Export to Endnote|
|Similar articles in Google Scholar|
|Similar articles in Baidu academic|
|Similar articles in Bing Scholar|
Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.