A Lens Inspection Method for High-Speed ​​and Accurate Stripping

Abstract

The invention discloses a lens detection method for high-speed accurate segmentation. The method comprises the following steps of: data analysis: for an SDI (Source Data Information) source, extracting YUV (Luma and Chroma) data from an SDI signal, and for a non-SDI source, carrying out decoding or signal separation to obtain the YUV or RGB (Red, Green, Blue) image information; lens detection: carrying out lens detection analysis on an input YUV or RGB image sequence, thus obtaining a detection result; assistant locating: locating a to-be-segmented file according to the lens detection result; and program segmentation: segmenting the located material. The method supports common variety of YUV and RGB formats, supports synchronization of acquisition and segmentation, and is free from recoding or format conversion, so that efficiency is improved; by utilization of small-sized images, the detection efficiency is improved, and the timeliness of the segmentation work is guaranteed; a grey level histogram difference and a gradient histogram difference are used comprehensively, and the detection accuracy is improved; by a gradient histogram, an adaptive threshold and a tolerance judging method, the detection accuracy is improved, and the influence of frequent flashing of news materials on the detection result is weakened.

Description

technical field [0001] The invention relates to the technical fields of video lens detection and video editing, in particular to a lens detection method for high-speed and accurate strip removal. Background technique [0002] In recent years, under the general trend of the development of new media dominated by the Internet and mobile media, various TV stations and content providers have used the Internet as the carrier of program dissemination to develop Internet audio-visual program services for TV stations. Through new carriers, new channels, and new Means to convey government orders to the society, publicize the principles and policies of the party and the country, report and publicize major political, economic, cultural and other activities or events in the city where they are located, and guide public opinion. With the integration of the three networks and the continuous development of new media services such as IPTV, online video, mobile TV, and mobile multimedia, the ...

AI Technical Summary

Problems solved by technology

[0007] However, there are many drawbacks in the existing lens detection method, which cannot fully meet the above-mentioned requirements of the stripping system: (1) The existing lens detection method supports a single format, only supports RGB data, does not support YUV signals, such as using RGB color Space or RGB to HSV color space to extract the detection method of feature value. In the working mode of SDI signal collection and stripping, the SDI signal records YUV information. The stripping system must build a YUV conversion for this lens detection method. The RGB module is time-consuming and labor-intensive and affects the efficiency of strip removal

(2) The efficiency of existing lens detection methods is low. For example, the detection method using edge feature information has a large amount of calculation, slow detection speed, and the detection efficiency is not even up to the decoding speed. A lot of precious time has to be spent waiting for the test results, and strip stripping is a time-sensitive work, and the serious lag of the test results will affect the effectiveness of strip stripping

(3) The detection accuracy is not high, especially for news videos of the main business of information services. There are frequent flashes in news materials, and titles and subtitles often fly in and out. Most lens detection methods are particularly sensitive to this, and it is easy to False alarms are generated, which reduces the accuracy of strip removal, and at the same time increases the workload of the staff virtually

Images