Implementation method of mosaic detection based on fixed video

Abstract

The invention relates to the field of image recognition, which discloses a mosaic detection realization method based on fixed video, which realizes automatic detection of video mosaic with high efficiency and accuracy. The method comprises the following steps of: a. preparing a preprocessed video, wherein the preprocessed video is that a time stamp is added to each frame image of the video; setting a two-dimensional code for each second video so that each frame image in each second video has the same two-dimensional code; b. extracting the eigenvalues of each frame image in the video and storing the eigenvalues to form a standard feature library; c. when that mosaic detection is carry out,playing the video by the equipment to be tested, and the video image is captured in real time by a computer for image acquisition; d. calculating the eigenvalues of the captured images and comparing the eigenvalues with the eigenvalues in the standard feature library to judge whether the video has mosaic or not.

Description

technical field [0001] The invention relates to the field of image recognition, in particular to a method for realizing mosaic detection based on fixed video. Background technique [0002] In traditional set-top boxes, the main basis for TUNER performance testing is video mosaic. In OTT and IPTV set-top boxes, video playback mosaic is an important test basis for checking the memory performance and software reliability of set-top boxes. Therefore, video mosaic detection is very important. However, due to the random timing and irregular graphics of the mosaic, especially for set-top boxes with post-processing functions, the shape of the mosaic after image restoration is more complicated. As a result, automated mosaic detection has always been a pain point and difficulty in the industry. [0003] Researchers have used the edge detection method of straight lines and rectangles to judge mosaics, but due to the limitations of edge detection itself, regular images such as straigh...

AI Technical Summary

Problems solved by technology

[0003] Researchers have used the edge detection method of straight lines and rectangles to judge mosaics, but due to the limitations of edge detection itself, regular images such as straight lines, chessboards, and windows in video content are easy to be misjudged. Mosaic is easy to be missed

There are also mosaic feature pre-analysis and deep learning methods to classify and learn mosaic feature values, but due to the complexity of the video image and the final image formed after the mosaic is superimposed, the final image is more complex, so the features learned in advance are completely incompatible.

[0004] Therefore, at present, the video mosaic test is basically performed by manual judgment in the industry, which has low efficiency, high cost, and high missed and false detection rates.

Images