基于多视点单目视觉的关联运动目标位姿高精度测量理论和方法研究

项目名称： 基于多视点单目视觉的关联运动目标位姿高精度测量理论和方法研究

项目编号： No.51275169

项目类型： 面上项目

立项/批准年度： 2013

项目学科： 机械、仪表工业

项目作者： 赵前程

作者单位： 湖南科技大学

项目金额： 80万元

中文摘要： 机器视觉测量应用广泛，是测量领域研究热点之一。本项目围绕多视点单目视觉关联运动目标位姿高精度测量中的"相机内部参数标定中线性参数和非线性参数的解耦"和"关联运动约束目标位姿测量的误差自补偿"两个关键科学问题，以"高精度"为目标、"误差控制"为主线，基于"引入约束关系建模以提高模型参数优化估计的鲁棒性和估计精度"的思想，从靶标设计、角点提取、离群角点判识、相机标定、位姿测量等主要环节入手，通过理论分析、数值仿真和实验研究开展相关基础理论和方法研究。主要研究内容包括：测量系统误差传递关系建模；基于局域灰度特征模型的角点高精度提取及其精度评价方法；基于几何关系约束的单相机内部参数和多相机外部姿态标定，重点是相机内部线性参数和非线性参数标定中的解耦机制；基于关联运动约束目标位姿测量的误差自补偿机理等。项目研究将丰富和发展相关的测量理论和方法，为基于单目视觉的位姿高精度测量应用提供理论和技术支撑。

中文关键词： 机器视觉；图像处理；相机标定；位姿测量；四轮定位

英文摘要： As one of the hotspots in measurement research field, machine vision measuring technology has been used widely. This research is aiming at the two key scientific problems in high accuracy measurement of correlative movement objective pose with monocular machine vision of multiple views, one is "Decoupling of linear parameters and nonlinear parameters in camera internal parameters calibration process", and the other is "Error self-compensation with correlative movement constraints in objective pose measurement process". In order to solve the two key scientific problems, this project will develop "error control" research in the whole process from planar target design, corner outliers detecting, camera calibration to pose measurement, etc. Based on the thought of "introducing constraint relation in the model to improve optimization algorithm robustness and model parameters estimation accuracy", this project will do the related basic theory and application technology research work by theoretical analysis, numerical simulation and experimental research. The main research includes: error transfer relation model of the measurement system, corner extraction method based on local area gray feature model and its accuracy evaluation, decoupling mechanism of linear parameters and nonlinear parameters in camera internal p

英文关键词： Machine Vision；Image Processing；Camera Calibration；Pose Measurement；Four-wheel Alignment