| With the rapid development of communication technology,video conferencing has been widely used in many industries due to its strong realtime and rich expression.During the pandemic,the demand for video conferencing increased,and screen sharing,which is used for collaborative problem-solving,document display,and operation demonstration,among other things,is a very important part of video conferencing systems.In the screen sharing scenario of video conferencing,the shared content is often slides,web pages,etc.,which are accompanied by introductions,resulting in a large number of redundant frames in the shared content when the user is not performing any operation.Moreover,in order to ensure the clarity of text and graphics during screen sharing,a high resolution is often used,which puts pressure on the user’s upstream bandwidth.However,the existing encoding process for screen content does not pre-judge a large number of redundant and repeated frames,but still performs frame-byframe encoding and transmission,resulting in wasted computing resources and bandwidth.This paper will optimize and improve the above scenarios with a large number of redundant duplicate frames in screen sharing.Aiming at the problem of identifying duplicate frames in screen sharing scene,this paper firstly designs a duplicate frame detection scheme based on image similarity measurement.This scheme uses grid parallelization to calculate the similarity of high resolution screen images,and achieves the beneficial effect of reducing time and improving accuracy.At the same time,an image similarity measurement algorithm combining two kinds of image features is proposed.The new algorithm is compared with many mainstream image similarity measurement algorithms.The results show that the new algorithm improves the accuracy and robustness of detection results,and meets the real-time requirements.In addition,the low probability of repeating frames is predicted by identifying whether the screen image contains natural images before the detection of repeating frames.Afterward,this article designs and implements a real-time screen sharing system based on variable frame rate encoding and integrates a duplicate frame detection scheme.By analyzing the screen in real-time,the system determines whether there are changes or small changes in the image to judge whether there are duplicate frames,and then transmits the screen through variable frame rate encoding,effectively reducing bandwidth and computational resource waste,relieving the pressure on the user’s upstream bandwidth and improving the quality and stability of video conferencing.Additionally,this article proposes a first-frame delay optimization scheme to ensure and optimize the user’s viewing experience after removing redundant duplicate frames.Finally,the entire system is tested,and the results show that the design and implementation of this system provide an effective method for optimizing real-time screen sharing systems,which has practicality and promotional value. |