阅读说明：本技术 一种筛选局部特征点的方法及装置 (Method and device for screening local feature points ) 是由 林天鹏 占云龙 左韶军 赵强 王林召 于 2018-06-13 设计创作，主要内容包括：本申请实施例提供一种筛选局部特征点的方法及装置,涉及图像处理领域,用于解决现有技术中筛选出的局部特征点在图像中分布比较集中且误差较大的问题。该方法包括：确定第一帧图像中每个网格所包含的局部特征点；利用第二帧图像中每个网格的局部特征点的信息分别对第一帧图像中处于相同位置的网格所包含的局部特征点进行筛选,以得到第一帧图像中每个网格保留的局部特征点,其中,第二帧图像为第一帧图像的上一帧图像,第一帧图像和第二帧图像的尺寸相同,第一帧图像和第二帧图像中处于相同位置的网格一一对应；根据第一帧图像中每个网格中保留的局部特征点的信息确定待输出的局部特征点。(The embodiment of the application provides a method and a device for screening local feature points, relates to the field of image processing, and is used for solving the problems that the local feature points screened in the prior art are distributed in an image more intensively and have larger errors. The method comprises the following steps: determining local feature points contained in each grid in the first frame image; respectively screening local feature points contained in grids at the same position in a first frame image by using the information of the local feature points of each grid in a second frame image to obtain local feature points reserved in each grid in the first frame image, wherein the second frame image is a previous frame image of the first frame image, the first frame image and the second frame image have the same size, and the grids at the same position in the first frame image and the second frame image are in one-to-one correspondence; and determining local feature points to be output according to the information of the local feature points reserved in each grid in the first frame of image.)

A method of screening local feature points, comprising:

determining local feature points contained in each grid in the first frame image;

respectively screening local feature points contained in grids at the same position in a first frame image by using information of the local feature points of each grid in a second frame image to obtain local feature points reserved by each grid in the first frame image, wherein the second frame image is a previous frame image of the first frame image, the first frame image and the second frame image have the same size, and the grids at the same position in the first frame image and the second frame image are in one-to-one correspondence;

and determining local feature points to be output according to the information of the local feature points reserved in each grid in the first frame of image.

The method for screening local feature points according to claim 1, wherein the determining the local feature points included in each grid in the first frame image comprises:

extracting a plurality of local feature points of the first frame image;

and determining the grid to which each local feature point in the plurality of local feature points belongs to obtain the local feature points contained in each grid in the first frame image.

The method for screening local feature points according to claim 1 or 2, wherein the screening the local feature points of the grids at the same position in the first frame image respectively by using the information of the local feature point of each grid in the second frame image to obtain the local feature point reserved for each grid in the first frame image comprises:

and respectively screening the local feature points contained in the grids at the same position in the first frame image by using the maximum response fraction value of the local feature point of each grid in the second frame image to obtain the local feature points reserved by each grid in the first frame image.

The method of claim 3, wherein the step of separately screening the local feature points included in the grids at the same position in the first frame image by using the maximum response score value of the local feature point of each grid in the second frame image comprises:

and comparing the response score value of the local feature point of each grid in the first frame image with the maximum response score value of the local feature point in the grid at the same position in the second frame image, reserving the local feature point more than or equal to the maximum response score value, and deleting the local feature point less than the maximum response score value.

The method for screening local feature points according to any one of claims 1 to 4, wherein the determining the local feature points to be output according to the information of the local feature points retained by each grid in the first frame image comprises:

and according to the sum of the number of the local feature points reserved in each grid in the first frame image and the response score value, sequencing a plurality of grids included in the first frame image, and determining the local feature points reserved in K grids in the front sequencing as the local feature points to be output.

The method of screening local feature points of claim 5, further comprising:

deleting information of local feature points of meshes other than the K meshes from the plurality of meshes included in the first frame image.

An apparatus for screening local feature points, the apparatus comprising:

the arithmetic unit is used for determining local feature points contained in each grid in the first frame image;

the arithmetic unit is further configured to respectively screen local feature points included in grids at the same position in the first frame image by using information of the local feature points of each grid in a second frame image, so as to obtain local feature points reserved in each grid in the first frame image, where the second frame image is a previous frame image of the first frame image, the first frame image and the second frame image have the same size, and the grids at the same position in the first frame image and the second frame image correspond to each other one to one;

the arithmetic unit is further configured to determine a local feature point to be output according to information of the local feature point retained in each grid in the first frame image;

a first buffer, configured to store information of a local feature point of each grid in the second frame image and information of a local feature point reserved in each grid in the first frame image;

and the second buffer is used for storing the local feature points reserved by each grid in the first frame image.

The apparatus for screening local feature points according to claim 7, wherein the operator is further configured to:

extracting a plurality of local feature points of the first frame image;

and determining the grid to which each local feature point in the plurality of local feature points belongs to obtain the local feature points contained in each grid in the first frame image.

The apparatus for screening local feature points according to claim 7 or 8, wherein the operator is further configured to:

and respectively screening the local feature points contained in the grids at the same position in the first frame image by using the maximum response fraction value of the local feature point of each grid in the second frame image to obtain the local feature points reserved by each grid in the first frame image.

The apparatus for screening local feature points according to claim 9, wherein the operator is further configured to:

and comparing the response score value of the local feature point of each grid in the first frame image with the maximum response score value of the local feature point in the grid at the same position in the second frame image, reserving the local feature point more than or equal to the maximum response score value, and deleting the local feature point less than the maximum response score value.

The apparatus for screening local feature points according to any one of claims 7 to 10, wherein the operator is further configured to:

and according to the sum of the number of the local feature points reserved in each grid in the first frame image and the response score value, sequencing a plurality of grids included in the first frame image, and determining the local feature points reserved in K grids in the front sequencing as the local feature points to be output.

The apparatus for screening local feature points according to claim 11, wherein the operator is further configured to:

deleting, from the first buffer, information of local feature points retained in meshes other than the K meshes among the meshes included in the first frame image.