基于改进网格划分统计的特征点快速匹配方法

Fast Feature Point Matching Algorithm Based On Improved Meshing Statistics

摘要:

针对图像特征点匹配算法大多存在数据量大和计算耗时长等问题，提出一种改进网格划分统计的特征点快速匹配算法。首先将图像的长宽比作为约束项，把图像划分成多个非重叠的方形状网格，并统计网格内的粗匹配特征点数量，然后利用改进的五宫格统计方法剔除错误匹配，即将特征点所在网格的相邻对称的四个网格作为邻域范围，把五宫格特征分数与新提出的阈值公式计算的值进行比较，最终得到精匹配特征点集；在OxFord数据集和实际拍摄的无人机遥感图像上，将本文算法与多种算法进行比较，实验结果表明，该方法在保证精确率和召回率接近当前最新的特征点快速匹配算法的情况下，运行速度相对提高了35.6 %，证明了特征点匹配的实时性和有效性。

Abstract:

Aiming at the problems of image feature point matching algorithm, such as large data volume and long calculation time, a fast feature matching algorithm for improved mesh segmentation statistics is proposed. Firstly, the aspect ratio of the image is taken as the constraint, the image is divided into a plurality of non-overlapping square shape meshes, and the number of rough matching feature points in the grid is counted, and then the modified five-square grid statistical method is used to eliminate the false match. The four adjacent grids of the grid where the feature points are located are taken as the neighborhood range, and the five-square grid feature score is compared with the value calculated by the newly proposed threshold formula, and finally the fine-matched feature point set is obtained; in the OxFord dataset Compared with the actual UAV remote sensing images, the algorithm is compared with various algorithms. The experimental results show that the proposed method can guarantee the accuracy and recall rate close to the current feature point fast matching algorithm. It has increased by 35.6 %, which proves the real-time and effectiveness of feature point matching.