Compressed domain video lens mutation and gradient union automatic segmentation method and system

Abstract

The invention relates to compressed domain video lens mutation and gradient union automatic segmentation method and system. In the invention, aiming at three factors of feature extraction, dissimilarity measure and segmentation rules affecting segmentation performance of lens, a video feature extraction method based on principal component analysis and a method for building a texture feature map with a compressed domain are provided; secondly, a dissimilarity measure method in a time domain with multi-scale is provided; a method based on adaptive threshold is provided according to parameters oflocal features which can effectively characterize lens change and adaptively determine the length spaced by an N-frame time domain through a dissimilarity measure operator spaced by a 1-frame time domain; and the dissimilarity measure operator and effective distinguishing rules are designed for mutation and gradient lens. The invention can effectively strengthen the robustness of interference ofa lens segmentation algorithm to a camera or object movement in the lens by fully considering the three factors affecting the lens segmentation performance, have strong noise resistant performance andlow false drop rate, be fast and accurate, and greatly enhance the segmentation performance of the lens.

Description

the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine

Login to View More

AI Technical Summary

Benefits of technology

This patented method improves image processing by taking into account various factors such as lighting conditions (lenses) movement, shutter speed, sensor resolutions, etc., which may cause issues when analyzed for smoothly changing images captured on cameras due to changes caused by external sources like vibrations during use. By doing this, it provides an effective way to accurately detect abrupt movements between different types of objects without being affected by these environmental disturbances.

Problems solved by technology

This patents discusses various technical means related to improving video shooting quality through automatic detection and annotation of videos without completely decoding each frame separately. Current solutions either rely on partial decomposition or compressible domains, leading to poor results over longer periods of playback. Additionally, existing approaches based solely on specific characteristics may result in less accurate and reliable video segments despite being able to accurately identify individual frames from multiple ones simultaneously.

Images