Object similarity comparison method and retrieval method and system based on video

Abstract

The invention relates to the field of video monitoring, and discloses an object similarity comparison method based on a video. The method comprises the following steps of respectively acquiring representative reference points of a first object and a second object from an image set including the first object and the second object, and computing a similarity between the objects according to the representative reference points of the first and second objects, wherein the representative reference point of each object is acquired by extracting a characteristic of the object from each frame including the object to serve as the characteristic of the object in a view angle; using the characteristic in each view angle as a reference point, and fitting a smooth hypercurve according to the reference points; and selecting the plurality of representative reference points, wherein a change rate of curvature between the adjacent representative reference points is greater than a predetermined threshold. According to the method provided by the invention, the similarity between the two objects is computed by selecting the reference points in the image with a low view angle overlapping degree as the representative reference points according to the change rate of the curvature of the reference points, so that computing efficiency and accuracy of a computing result are improved.

Description

technical field [0001] The invention relates to the field of video monitoring, in particular to comparison and retrieval of similarity of video screen objects. Background technique [0002] At present, there are many existing technologies similar to the present invention on the market, and these technologies more or less refer to key technical processes such as three-dimensional model retrieval and multi-feature fusion. [0003] CN101281545A proposes a three-dimensional model retrieval method based on multi-feature correlation feedback, in which it is mentioned that the server side processes each three-dimensional model in the three-dimensional model database. [0004] CN101593205A proposes a video-based three-dimensional model retrieval method, in which it is mentioned that the three-dimensional model database is preprocessed to generate a two-dimensional contour database. [0005] CN103116904A proposes a two-dimensional feature extraction system of a three-dimensional mod...

AI Technical Summary

Problems solved by technology

[0007] (1) The existing technology requires explicit 3D modeling of the objects to be retrieved, that is, the database needs to store a large number of 3D models of objects, which greatly increases the storage capacity, and on the other hand, requires a lot of energy to collect interested objects the model;

[0008] (2) Existing technologies mostly use the projected outline or gradient features of the 3D model on the 2D plane, and these artificially designed features cannot fully cover all the useful information on the surface of the object under the current viewing angle;

[0009] (3) When the existing technology calculates the similarity between two objects, it only calculates the measurement between the corresponding viewing angles of the two objects, without considering the correlation and importance between the various viewing angles of the objects, and the retrieval accuracy needs to be improved

Images