Image matching method and device based on feature detection, equipment and medium

Abstract

The invention provides an image matching method and device based on feature detection, equipment and a medium, and the method comprises the steps: inputting an original image and a template image, detecting feature points through an SIFT method, extracting N feature points, and generating a feature descriptor according to the coordinate, scale and direction of each feature point; wherein each image generates a target feature point set; after a target feature point set is generated, according to the feature point matching degree of each template image and the original image, extracting a plurality of previous images with the highest matching score by using SIFT and BoW algorithms; carrying out local feature matching by SIFT and AdaLAM algorithms, and selecting an optimal matching image from the plurality of images with the highest matching scores; and outputting a matching result. The image matching method based on feature detection provided by the invention is good in matching effect and can efficiently and quickly filter out the outer points.

Description

technical field [0001] The present invention relates to the technical field of image matching models and algorithms of computer vision, in particular to an image matching method, device, equipment and medium for feature detection. Background technique [0002] Image matching is a basic problem in computer vision. Accurate and efficient image matching algorithms provide a solid underlying foundation for solving other problems. Image matching methods based on SIFT features can be widely used in industries such as industrial inspection, indoor navigation, and security monitoring. Most of the methods based on SIFT features rely on the BoW model. BoW technology was originally applied to the document retrieval system, which is characterized by the need for a large number of words to generate codebooks, and one of the characteristics of SIFT feature extraction is the large number, even if the image contains only a few objects, it can also generate many SIFT features, Use the SIFT ...

AI Technical Summary

Problems solved by technology

[0003] However, the traditional SIFT algorithm plus BoW model has the following problems: SIFT algorithm plus BoW model is a global feature matching based on deep learning, which requires the creation of a data set and model training, and the traditional SIFT algorithm extracts image features. , is greatly affected by illumination; in addition, in the image matching task, there are many outliers in the initial matching, and it is currently difficult to filter the outliers efficiently and quickly

Images