Real-time Key Frame Selection Method in Mine Wireless Distributed Video Coding

Abstract

The present invention relates to a method for the real-time selection of key frames when mining wireless distributed video coding, including the steps: constructing an incremental curve of inter-frame features: acquiring a video sequence, and extracting the global and local features of the images; using the colour features of the images as an indicator to measure the global features of the images; using feature points as local features of the images; and on the basis of the global features and the local features of the images, calculating the overall feature similarity of the images; on the basis of the overall feature similarity of the images, constructing an incremental curve of the inter-frame feature differences: using the frame number as the abscissa and the cumulative value of the inter-frame overall feature similarity of the images as the ordinate, plotting an incremental curve of the inter-frame feature differences of the images; and after obtaining the incremental curve of the inter-frame feature differences, using a key frame real-time selection algorithm to implement rapid adaptive selection of key frames. The present invention enables the rapid and low-complexity reflection of dynamic changes in monitored image content, effectively solving the problem of real-time selection of key frames when mining wireless distributed video coding.

Description

technical field [0001] The invention relates to a method for real-time selection of key frames, in particular to a method for real-time selection of key frames in mine wireless distributed video coding. Background technique [0002] With the increase of mine depth, coal mine disasters are becoming more and more serious, and new requirements are put forward for the safety guarantee level and emergency response capacity of coal mines. The wireless video sensor network uses a large number of energy-limited micro-nodes to collect, transmit and process real-time image information of the mine environment, so that ground experts and decision-makers can understand the rescue scene in real time and make scientific decisions and commands. In the wireless video sensor network, distributed video coding adopts "intra-frame coding + inter-frame decoding" technology, and "inter-frame predictive decoding" is performed by exploring the correlation of video signals at the decoding end, thereb...

AI Technical Summary

Problems solved by technology

Among them, for the video content with large content changes and violent motion, the method based on shot segmentation, a single image frame cannot fully and effectively describe the content of the shot, and some important content in the video will be lost; the method based on image content analysis will overselect Many key frames cause redundancy. In addition, the selection of key frame screening thresholds and the selection of underlying features also have an impact on the selection performance of key frames; methods based on motion analysis require a very large amount of calculation when calculating the amount of motion in the video. , in addition, the selection of key frames based on local information is not robust; based on the clustering method, the image frames are divided into different categories according to a certain degree of similarity, so that the selected key frames do not retain the order of time Sequence, resulting in the inability to understand the dynamic information of the video by observing key frames, and the application in distributed video coding is limited

It can be seen that in the current distributed video coding, in order to improve the real-time performance of the coding, the key frame selection is mainly through the method of periodically selecting key frames, without considering the correlation between frames, which is not conducive to mine bandwidth limited, high High-efficiency video coding in a noisy environment is mainly manifested in the following aspects: for the relatively smooth motion part of the video sequence, too many key frames are selected, which consumes a large amount of mine wireless network resources and brings a huge processing burden to the transmission node; In the case of violent motion or sudden changes in the picture, periodic selection of key frames is not enough to generate high-quality side information frames, which reduces the quality of image reconstruction at the decoding end.

Images