无人驾驶车辆基于角点和斑点的特征提取算法

针对运行在计算资源有限的车载嵌入式系统中的视觉里程计算法实时性较差的问题,提出一种基于Harris和SIFT相结合的图像匹配方法——Harris-SIFT算法。在介绍了SIFT算法的基础上,给出了Harris-SIFT算法的原理:使用Harris算法提取图像中的角点作为特征候选点,再利用SIFT算法在Harris的特征候选点中进行特征点提取。通过实例用Matlab软件对算法进行了仿真,并对算法的复杂度及各种性能进行了分析。结果表明,所提出的方法在特征检测模块中降低了算法的运算量、提高了特征点提取速度。Harris-SIFT算法可用于实时视觉里程计系统中,进而可使视觉里程计在车载嵌入式导航系统上得到广泛的应用。

A feature extraction algorithm based on corner and spots in self-driving vehicles

To solve the poor real-time performance problem of the visual odometry based on embedded system with limited computing resources, an image matching method based on Harris and SIFT is proposed, namely the Harris-SIFT algorithm. On the basis of the review of SIFT algorithm, the principle of Harris-SIFT algorithm is provided. First, Harris algorithm is used to extract the corners of the image as candidate feature points, and scale invariant feature transform (SIFT) features are extracted from those candidate feature points. At last, through an example, the algorithm is simulated by Matlab, then the complexity and other performance of the algorithm are analyzed. The experimental results show that the proposed method reduces the computational complexity and improves the speed of feature extraction. Harris-SIFT algorithm can be used in the real-time vision odometer system, and will bring about a wide application of visual odometry in embedded navigation system.