CAJ | 학술논문

流媒体直播点播系统中，将实时采集的音视频流封装为流媒体文件是一项必不可少的工作。目前，FFmpeg封装音视频流广泛采用先解码、后编码封装的方式，针对该方式内存占用较多、CPU使用率较高的情况，提出将实时的 H.264视频流和AAC音频流不解码直接封装为MP4文件的实现方法。该方法基于FFmpeg SDK进行二次开发，可处理实时采集的H.264视频流和AAC音频流。实验结果表明，该方法封装实时音视频流占用内存少，CPU平均使用率较低，适合在流媒体直播点播系统中使用。
류매체직파점파계통중，장실시채집적음시빈류봉장위류매체문건시일항필불가소적공작。목전，FFmpeg봉장음시빈류엄범채용선해마、후편마봉장적방식，침대해방식내존점용교다、CPU사용솔교고적정황，제출장실시적 H.264시빈류화AAC음빈류불해마직접봉장위MP4문건적실현방법。해방법기우FFmpeg SDK진행이차개발，가처리실시채집적H.264시빈류화AAC음빈류。실험결과표명，해방법봉장실시음시빈류점용내존소，CPU평균사용솔교저，괄합재류매체직파점파계통중사용。
In the VOD and live streaming system, encapsulating the video and audio stream into a container is necessary. Analyzes the decoding and encoding method to pack streams using FFmpeg, in order to solve high usage rate of memory and CPU, proposes a novel method of pack-ing streams without decoding into MP4 container. Based on FFmpeg SDK, packs H.264 video streams and AAC audio streams into MP4 container without decoding. Experimental results show that this method reduces the usage rate of memory and CPU, suitable used in the VOD and live stream systems.