Video shot segmentation method and system

Abstract

The invention discloses a video shot segmentation method and system. The method comprises the steps of obtaining all images in a video and converting the images into HSV images; partitioning the HSV image into blocks; calculating the color histogram difference of the blocks of two adjacent frames of images on an HSV single channel; calculating the average color difference of the blocks of two adjacent frames of images on the HSV channel; calculating the color histogram difference of two adjacent frames of images; judging whether the color histogram difference of the two adjacent frames of images is greater than or equal to a set abrupt change threshold value or not, if the color histogram difference of the two adjacent frames of images is greater than or equal to the set abrupt change threshold value, judging as an abrupt change lens, and segmenting the previous frame of image and the next frame of image of the abrupt change lens; and if the color histogram difference between the two adjacent frames of images is greater than or equal to a set gradual change threshold value, determining that the shot is a gradual change shot, and continuously detecting the remaining images until allframes in the video are detected. Accuracy of detecting the abrupt change lens and the gradual change lens can be improved at the same time.

Description

technical field [0001] The present invention relates to the field of image processing, in particular to a video shot segmentation method and system Background technique [0002] The task of shot segmentation is to divide a complete video into fragments in units of shots. There are two types of sudden shots and gradual shots: sudden shots mean that there is no obvious transition between two discontinuous frames; and gradual shots refer to The most important thing is that there are transition forms such as fade in, fade out, and dissolve between two discontinuous frames. The common shot segmentation method is mainly based on the difference value between two adjacent frames for discrimination. On this basis, various shot segmentation methods have evolved, such as using multi-feature fusion methods for video shot segmentation, using matrix cosine similarity methods, frame transition parameters and frame estimation errors based on local features, and shot segmentation algorithms...

AI Technical Summary

Problems solved by technology

These methods have high detection accuracy for shot cuts in different scenes, but poor detection results for shot cuts in the same scene

In addition, the detection accuracy of the above method for progressive lens is also low, so it is necessary to improve the accuracy of lens segmentation

Images