Video data linear deviation main feature extraction device and method

Abstract

The present invention provides a video data linear deviation main feature extraction device and method. The device includes a video receiving module, a linear deviation calculation module, a video data module and a feature extraction module; the video receiving module is used for importing video data; the linear deviation calculation module is used for obtaining linear deviations corresponding tothe imported video data according to the imported video data and combining the linear deviations; the video data module is used for storing data, wherein the data comprise at least video data, lineardeviations and hidden features; and the feature extraction module is used for extracting the hidden features of the video data according to a linear deviation combination. According to the video datalinear deviation main feature extraction device and method of the present invention, the hidden features of the video data are extracted; the most suitable linear deviation combination is selected according to the different video data, so that the most accurate decomposition result can be obtained; and therefore, potential laws and meanings contained in the video data can be disclosed, and the problem of the extraction of massive video data in large-scale and complex systems can be solved.

Description

technical field [0001] The invention relates to the field of computer image processing, in particular to a device and method for extracting main features of linear deviation of video data. Background technique [0002] Video data feature extraction, as one of the most important links in computer image processing, has always been the main problem and research hotspot in computer science and engineering. Since video data can be converted into high-dimensional sparse data matrix, matrix decomposition technology can be used to obtain Extracting latent features from the specified video data to reveal the underlying rules and meanings contained in the video data is a major technology for feature extraction. [0003] At present, many existing technical methods are based on the pixel domain. Among them, the matrix decomposition technology has been able to combine linear deviations and extract latent features from video data. But at present, linear deviation is divided into two cate...

AI Technical Summary

Problems solved by technology

Different video data combined with different linear deviation combinations may improve the accuracy of latent feature extraction or reduce the accuracy of latent feature extraction, and may even lead to the inability to effectively extract latent features.

This is because most of the existing matrix factorization techniques use complete linear deviations for latent feature extraction, and it is impossible to select the most suitable combination of linear deviations for different video data to obtain the most accurate decomposition results.

Images