面向视觉SLAM的快速均匀特征点提取方法CSCD

针对现有特征点提取方法中特征点提取结果冗余且均匀性差,导致SLAM定位精度及计算效率低的问题,提出了一种面向视觉SLAM的快速均匀特征点提取方法。在该算法的特征点提取部分,创新性地根据目标特征点提取数量对图像进行网格划分,使得特征点均匀分布在整幅图像中。同时,提出了一种兼顾全局和局部信息的均匀性评价标准,更符合视觉SLAM对特征点分布的要求。在NVIDIA Jetson TX2嵌入式平台上进行了实验验证,实验结果表明,新提出的特征点提取方法降低了特征信息冗余及特征信息缺失的概率,提升了SLAM定位精度,在计算资源有限的平台上具有实际应用价值。

Fast Uniform Feature Point Extraction Method for Visual SLAM

Aiming at the problem of redundant and poor uniformity of feature point extraction results in the existing feature point extraction methods, which leads to the low accuracy and computational efficiency of SLAM pose estimation, a fast uniform feature point extraction method for visual SLAM is proposed. In the feature point extraction part of the proposed algorithm, the image is creatively meshed according to the number of target feature points, so that the feature points are evenly distributed in the whole image. At the same time, a uniformity evaluation standard considering both global and local information is proposed, and the new method is more in line with the requirements of visual SLAM for feature point distribution. Experimental verification is carried out on NVIDIA Jetson TX2 embedded platform. The experimental results show that the newly proposed feature point extraction method reduces the probability of feature information redundancy and missing, improves the accuracy of LAM pose estimation, and has practical application value on the platform with limited computing resources.