Feature matching method and device
A feature matching and feature point technology, applied in the field of computer vision, can solve the problems of large amount of feature calculation, difficult to apply the speed of the scene, time-consuming and other problems, achieve low noise impact, improve feature extraction efficiency, and accelerate model training Effect
Inactive Publication Date: 2019-11-26
BEIJING YINGPU TECH CO LTD
View PDF9 Cites 1 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0005] In the above-mentioned related feature matching technology, due to the huge amount of feature calculation, the process of feature point extraction is extremely time-consuming, so it is difficult to apply to scenes that focus on speed, and when using the matching results to achieve image stitching or object tracking, Matching points need to be further screened to obtain excellent matching points
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0036] figure 1 is a flowchart of a feature matching method according to an embodiment of the present application. Such as figure 1 As shown, the feature matching method may generally include the following steps:
[0037] Step 101, input the first picture to be subjected to feature matching into a pre-trained first convolutional neural network (CNN), wherein the first CNN network has a network architecture similar to that of the computer vision group (VGG).
[0038] In a specific application, before step 101 is executed, a structure similar to VGG can be selected to construct the first CNN and the second CNN, and then the first CNN and the second CNN can be trained so that the first CNN can detect the input image Stable feature points in the middle, map the input picture to a picture output with the same resolution but with feature point identification, the second CNN can perform feature matching on the input pair of pictures with feature point identification, and generate t...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The invention discloses a feature matching method and device. The method comprises the following steps: successively inputting a first picture and a second picture to be subjected to feature matchinginto a pre-trained first CNN; enabling the first CNN to identify feature points of the input first picture and the input second picture respectively, identify the identified feature points on the first picture and the second picture, and output the first picture and the second picture with feature point identifiers respectively; inputting the first picture with the feature point identifier and thesecond picture with the feature point identifier into a pre-trained second CNN; and enabling the second CNN to perform feature matching on the first picture with the feature point identifier and thesecond picture with the feature point identifier to obtain a homography matrix between the feature points of the first picture and the feature points of the second picture so as to obtain a corresponding relationship between the feature points of the first picture and the feature points of the second picture.
Description
technical field [0001] The present application relates to the field of computer vision, in particular to a feature matching method and device. Background technique [0002] Each image has unique pixel points, which can be considered as features of the image and can be called feature points. At present, the concept of interest points (also known as key points or feature points) has been widely used. In the field of computer vision, image feature matching is based on feature points. Therefore, how to define and find out the The feature points are very important for the subsequent feature matching. Select some feature points from the image and analyze the image locally instead of observing the entire image, as long as there are enough detectable interest points in the image, and these interest points are different and stable, and can be accurately located , then these feature points are very effective in the fields of target recognition, image registration, visual tracking, a...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More IPC IPC(8): G06K9/46G06K9/62G06N3/04G06N3/08
CPCG06N3/08G06V10/462G06N3/045G06F18/22G06F18/214
Inventor 彭浩
Owner BEIJING YINGPU TECH CO LTD