论文部分内容阅读
剖面匹配在道路提取中得到了广泛的应用,但现有的相关文献中普遍利用数学模型模拟道路轨迹、利用最小二乘剖面匹配的结果进一步修正数学模型预测的道路点的方法,这种方法与实际的路况有较大的差别。本文中,用户初始化后获得道路的起始点、前进方向、道路的宽度和模板剖面。通过将模板剖面沿道路前进方向平移和旋转生成一系列目标剖面,求目标剖面与模板剖面的灰度差的平方和,在求平方和时给剖面的道路部分更多权重,最小的平方和对应的目标剖面的中点为最精确的道路点,迭代上述步骤追踪道路轨迹。经试验证明,改进的剖面匹配算子是一种稳健高效的道路追踪算法。
Profile matching has been widely used in road extraction. However, in the existing literature, mathematical models are generally used to simulate the road trajectory, and the results of the least squares cross-section matching are used to further correct the road points predicted by the mathematical model. The actual road conditions have a greater difference. In this paper, the user gets the initial point of the road, the direction of advance, the width of the road and the profile of the template after initialization. A series of target profiles are generated by translating and rotating the template profile along the direction of the road. The square sum of the gray difference between the target profile and the template profile is obtained. When the square sum is given, more weight is given to the road segment of the profile. The minimum square sum corresponds The midpoint of the target profile is the most accurate waypoint, tracking the path of the track iteratively. The experiment proves that the improved profile matching operator is a robust and efficient road tracking algorithm.