Image matching system based on ORB algorithm and matching method thereof

Abstract

The invention discloses an image matching system based on an ORB algorithm and a matching method thereof. The image matching system comprises an information acquisition platform, an information processing platform, an information analysis platform and an output platform, the information processing platform is electrically connected with the information acquisition platform, the information analysis platform is electrically connected with the information processing platform, and the output platform is electrically connected with the information analysis platform. According to the scheme, the information acquisition platform can capture a real scene in real time and process the frame number of the captured image in real time, a person does not need to walk on a fixed route, and the information processing platform can perform feature recognition on the real scene image, detect the content of the real scene image, filter junk information in the real scene image and process the image, so that the image becomes relatively clear, the gradient of the image can be integrated by the information analysis platform of the invention and can be calculated and compared, the intensity is accurate, and the output platform of the invention is convenient for multi-window contact, so that the conclusion can be conveniently exported to different platforms in real time.

Description

technical field [0001] The present application relates to the technical field of image matching, in particular, to an ORB algorithm-based image matching system and a matching method thereof. Background technique [0002] Currently, there are many methods for image matching, among which SIFT, SURF and ORB algorithms are the most widely used. Among them, the epoch-making significance is the scale-invariant feature transformation SIFT algorithm proposed by Lowe in 1999, and it was supplemented and improved in 2004. This algorithm is widely used and plays a huge role in the fields of target recognition, image stitching, 3D reconstruction and so on. The features detected by the SIFT algorithm are local features. These features are scale and rotation invariant. These features are very robust to brightness and noise, and can be correctly identified in the case of low probability mismatches. , are highly distinguishable. At the same time, the SIFT algorithm is divided into four s...

AI Technical Summary

Problems solved by technology

[0005] Now due to the large population base, it is difficult to control the flow of the population. Now the number of people in some activity places must be controlled to prevent exceeding the standard. The existing mainstream living detection schemes are divided into blocking type (light / wave blocking counting) , gate counting and infrared counting, the above three schemes have deficiencies;

[0006] 1. The counting inaccuracy of the blocking type is serious in a densely populated environment, and it is impossible to determine the entry and exit in a conventional office environment, but simply block the accumulation

[0007] 2. The gate counting type usually requires large-scale construction and is used for deployment at the entrance and exit of office buildings. Its cost is high and it cannot be deployed for offices due to its large size.

[0008] 3. Due to the high deployment cost and narrow judgment range, infrared counters require users to perform calculations and analysis after walking through fixed lines. They are mostly used in security environments such as airports and stations, and are not suitable for ordinary office scenarios.

Images