A method and system for detecting image similarity

Abstract

The invention relates to a method and system for detecting similarity of pictures. The detection method comprises: respectively screening out Internet picture feature points and reference picture feature points in Internet pictures and reference pictures; Obtain matching feature points in the Internet picture and the reference picture; when the number of matching feature points is greater than or equal to a preset threshold, the Internet picture and the reference picture are similar. The embodiment of the present invention obtains whether the feature points of the Internet picture and the reference picture match by acquiring the feature points of the two pictures for comparison, and counts the number of matching feature points. When the number of matching feature points exceeds the preset When the threshold is reached, it can be confirmed that the Internet picture is similar to the reference picture. The embodiment of the present invention has less calculation data and strong anti-noise ability, and realizes rapid identification and verification of whether pictures are similar.

Description

technical field [0001] The invention relates to the technical field of picture recognition, in particular to a picture similarity detection method and system. Background technique [0002] At present, Google's image search function uses a perceptual hashing algorithm to detect similar images. The principle is to generate a fingerprint string for each image and compare the distance between the fingerprints of different images. The shorter the distance, the more similar the image. The implementation of this algorithm is relatively simple and fast, but when a small content change is made to the picture, such as adding a watermark or text to the picture, it cannot be recognized, and there is a problem that the recognition of similar pictures is not stable enough. [0003] In actual work, in order to improve the accuracy of image similarity detection, a method based on feature point matching is often used to achieve fast image matching. The method of feature point matching has hi...

AI Technical Summary

Problems solved by technology

The more popular feature point detection algorithms mainly include the Harris algorithm, as well as the SIFT algorithm with a faster detection speed. The Harris algorithm has poor anti-noise ability, and the SIFT algorithm has a high dimension of feature vectors, which requires a large amount of calculation data, takes a long time, and cannot be realized real-time effect

[0004] Therefore, the existing image similarity calculation method has the problems of low execution efficiency and unstable recognition

Images