Live555实时视频流应用总结

1，linux 环境：
官网上下载，下载地址：http://www.live555.com/liveMedia/public/
live555 版本：“2018.12.14”
参考：http://www.live555.com/liveMedia/faq.html 这个FAQ要仔细阅读。
2,编译
根据不同的平台来配置，并生成对应的Makefile
2.1 ARM平台:
修改交叉编译工具
cp config.armlinux config.arm
vi config.arm
CROSS_COMPILE?= arm-buildroot-linux-uclibcgnueabi-
生成Makefile: ./genMakefiles arm
2.2 Linux 64位平台（x86-64 ）：
./genMakefiles linux-64bit
2.3 Linux 32位平台(x86)：
./genMakefiles linux
make

生成mediaServer/live555MediaServer

3,测试
3.1，mediaServer下会生成 live555MediaServer。
live555MediaServer test.264
如果出现Correct this by increasing “OutPacketBuffer::maxSize” to at least 186818, before creating this ‘RTPSink’. (Current value is 100000.)
在DynamicRTSPServer.cpp文件ServerMediaSession* createNewSMS（）
里修改OutPacketBuffer::maxSize

if (strcmp(extension, ".264") == 0) {// Assumed to be a H.264 Video Elementary Stream file:NEW_SMS("H.264 Video");OutPacketBuffer::maxSize = 300000; //100000;// allow for some possibly large H.264 framessms->addSubsession(H264VideoFileServerMediaSubsession::createNew(env, fileName, reuseSource));}

createNewSMS是在RTSP setup时调用的。
3.2，testProgs
testProgs 目录下各种测试文件,每个文件的作用和用法，官网上有详细的介绍。这些测试用例目前基本上都是以文件的形式作为输入源，下面重点介绍以实时流的形式作为输入源的2种方法。
主要是参考testH264VideoStreamer 和testOnDemandRTSPServer来修改。

4.不用读文件，使用实时视频流作为输入源
**
最简单的方法：将实时视频流推送到一个FIFO管道(或stdin),将文件名改为这个管道的文件名，这里不做详细介绍了。

4.1  方法1，参考testH264VideoStreamer

参考"liveMedia/DeviceSource.cpp"
定义一个H264LiveVideoSource例并继承DeviceSource，填充其成员，

void play() {// Open the input file as a 'byte-stream file source':ByteStreamFileSource* fileSource=ByteStreamFileSource::createNew(*env, inputFileName);
}

这里用H264LiveVideoSource代替ByteStreamFileSource

H264LiveVideoSource类后面会给出具体的代码。
修改testH264VideoStreamer.cpp main()

ServerMediaSession* sms= ServerMediaSession::createNew(*env, "testStream", NULL,"Session streamed by \"testH264VideoStreamer\"",True /*SSM*/);

修改play()函数如下：

void play() {// Open the input file as a 'byte-stream file source':#if 1H264LiveVideoSource* fileSource= new H264LiveVideoSource(*env);if (fileSource == NULL) {*env << "Unable to open file \"" << inputFileName<< "\" as a byte-stream file source\n";exit(1);}#elseByteStreamFileSource* fileSource= ByteStreamFileSource::createNew(*env, inputFileName);if (fileSource == NULL) {*env << "Unable to open file \"" << inputFileName<< "\" as a byte-stream file source\n";exit(1);}#endifFramedSource* videoES = fileSource;// Create a framer for the Video Elementary Stream:videoSource = H264VideoStreamFramer::createNew(*env, videoES);// Finally, start playing:*env << "Beginning to read from file...\n";videoSink->startPlaying(*videoSource, afterPlaying, videoSink);
}

4.2 方法2，参考testOnDemandRTSPServer
1）set the variable “reuseFirstSource” to “True”
2）根据类H264VideoFileServerMediaSubsession，新建一个新类H264LiveVideoServerMediaSubsession, implementation of the two pure virtual functions “createNewStreamSource()” and “createNewRTPSink()”
在createNewStreamSource()里用上面的H264LiveVideoSource代替ByteStreamFileSource。

H264VideoRTPSink继承关系：
H264VideoRTPSink->H264or5VideoRTPSink->VideoRTPSink->MultiFramedRTPSink->RTPSink->MediaSink->Medium。
H264VideoRTPSource继承关系：
H264VideoRTPSource->MultiFramedRTPSource->RTPSource->FramedSource->MediaSource->Medium.
H264VideoStreamFramer继承关系：
H264VideoStreamFramer->H264or5VideoStreamFramer->MPEGVideoStreamFramer->FramedFilter->FramedSource ->MediaSource->Medium.

下面列出具体实现的代码。

#ifndef _H264_LIVE_VIDEO_SERVER_MEDIA_SUBSESSION_HH
#define _H264_LIVE_VIDEO_SERVER_MEDIA_SUBSESSION_HH#include "OnDemandServerMediaSubsession.hh"
#include "liveMedia.hh"
#include "UsageEnvironment.hh"
#include "GroupsockHelper.hh"class H264LiveVideoServerMediaSubsession: public OnDemandServerMediaSubsession
{
public:H264LiveVideoServerMediaSubsession(UsageEnvironment & env,Boolean reuseFirstSource);~H264LiveVideoServerMediaSubsession();static H264LiveVideoServerMediaSubsession* createNew(UsageEnvironment& env,Boolean reuseFirstSource);
public: // new virtual functions, defined by all subclassesvirtual FramedSource* createNewStreamSource(unsigned clientSessionId,unsigned& estBitrate) ;// "estBitrate" is the stream's estimated bitrate, in kbpsvirtual RTPSink* createNewRTPSink(Groupsock* rtpGroupsock,unsigned char rtpPayloadTypeIfDynamic,FramedSource* inputSource);virtual char const * getAuxSDPLine(RTPSink * rtpSink, FramedSource * inputSource);static H264LiveVideoServerMediaSubsession* createNew(UsageEnvironment & env, FramedSource * source);static void afterPlayingDummy(void * ptr);static void chkForAuxSDPLine(void * ptr);void chkForAuxSDPLine1();
private:FramedSource * m_pSource;char * m_pSDPLine;RTPSink * m_pDummyRTPSink;char m_done;
};#endif

H264LiveVideoServerMediaSubsession.cpp

H264LiveVideoServerMediaSubsession::H264LiveVideoServerMediaSubsession(UsageEnvironment & env,Boolean reuseFirstSource):OnDemandServerMediaSubsession(env,reuseFirstSource)
{m_pSource = NULL;//source;m_pSDPLine = NULL;m_pDummyRTPSink =NULL;m_done=0;}H264LiveVideoServerMediaSubsession::~H264LiveVideoServerMediaSubsession()
{if (m_pSDPLine){free(m_pSDPLine);}
}
H264LiveVideoServerMediaSubsession * H264LiveVideoServerMediaSubsession::createNew(UsageEnvironment& env,Boolean reuseFirstSource)
{return new H264LiveVideoServerMediaSubsession(env,reuseFirstSource);
}
FramedSource * H264LiveVideoServerMediaSubsession::createNewStreamSource(unsigned clientSessionId, unsigned & estBitrate)
{//printf("===========createNewStreamSource===================\n");estBitrate = 500; return H264VideoStreamFramer::createNew(envir(), new H264LiveVideoSource(envir()));
}RTPSink * H264LiveVideoServerMediaSubsession::createNewRTPSink(Groupsock * rtpGroupsock, unsigned char rtpPayloadTypeIfDynamic, FramedSource * inputSource)
{return H264VideoRTPSink::createNew(envir(), rtpGroupsock, rtpPayloadTypeIfDynamic);
}char const *H264LiveVideoServerMediaSubsession::getAuxSDPLine(RTPSink * rtpSink, FramedSource * inputSource){if (m_pSDPLine){return m_pSDPLine;}m_pDummyRTPSink = rtpSink;if(NULL == m_pDummyRTPSink)return NULL;//mp_dummy_rtpsink->startPlaying(*source, afterPlayingDummy, this);m_pDummyRTPSink->startPlaying(*inputSource, 0, 0);chkForAuxSDPLine(this);m_done = 0;char const *dasl = m_pDummyRTPSink->auxSDPLine();envir().taskScheduler().doEventLoop(&m_done);if(dasl)m_pSDPLine = strdup(dasl);m_pDummyRTPSink->stopPlaying();return m_pSDPLine;}
void H264LiveVideoServerMediaSubsession::afterPlayingDummy(void * ptr)
{H264LiveVideoServerMediaSubsession * This = (H264LiveVideoServerMediaSubsession *)ptr;This->m_done = ~0;
}void H264LiveVideoServerMediaSubsession::chkForAuxSDPLine(void * ptr)
{H264LiveVideoServerMediaSubsession * This = (H264LiveVideoServerMediaSubsession *)ptr;This->chkForAuxSDPLine1();
}void H264LiveVideoServerMediaSubsession::chkForAuxSDPLine1()
{if (m_pDummyRTPSink->auxSDPLine()){m_done =  ~0;}else{double delay = 1000.0 / (FRAME_PER_SEC);  // msint to_delay = delay * 1000;  // usnextTask() = envir().taskScheduler().scheduleDelayedTask(to_delay, chkForAuxSDPLine, this);}
}

修改testOnDemandRTSPServer.cpp文件如下:
在main()里加入下面的代码

// A H.264 live video  stream:{OutPacketBuffer::maxSize = 300000;char const* streamName = "h264LiveVideo";char const* inputFileName = "test";ServerMediaSession* sms= ServerMediaSession::createNew(*env, streamName, streamName,descriptionString,True);UsageEnvironment& envr = rtspServer->envir();envr << "\n\"" << sms<< "\"\n" ;if(NULL == sms)printf("sms is null  \n");sms->addSubsession(H264LiveVideoServerMediaSubsession ::createNew(*env,True));rtspServer->addServerMediaSession(sms);announceStream(rtspServer, sms, streamName, inputFileName);}

H264LiveVideoSource.hh

#ifndef _H264_LIVE_VIDEO_SOURCE_HH
#define _H264_LIVE_VIDEO_SOURCE_HH#ifndef _FRAMED_SOURCE_HH
#include "FramedSource.hh"
#endif
#include "DeviceSource.hh"class H264LiveVideoSource: public FramedSource {
public:H264LiveVideoSource(UsageEnvironment& env);// called only by createNew()virtual ~H264LiveVideoSource();private:// redefined virtual functions:virtual void doGetNextFrame();//virtual void doStopGettingFrames();int maxFrameSize();static void getNextFrame(void * ptr);void GetFrameData();private:void *m_pToken;char *m_pFrameBuffer;char *fTruncatedBytes;int fTruncatedBytesNum;
};#endif

H264LiveVideoSource.cpp

#include "H264LiveVideoSource.hh"
//#include "InputFile.hh"
#include "GroupsockHelper.hh"#define FRAME_BUF_SIZE  (1024*1024)
#define FMAX (300000)
H264LiveVideoSource::H264LiveVideoSource(UsageEnvironment& env):FramedSource(env),
m_pToken(0),
m_pFrameBuffer(0)，fTruncatedBytesNum(0),fTruncatedBytes(0)
{m_pFrameBuffer = new char[FRAME_BUF_SIZE];fTruncatedBytes = new char[FRAME_BUF_SIZE];if(m_pFrameBuffer == NULL || fTruncatedBytes== NULL ){printf("[MEDIA SERVER] error malloc data buffer failed\n");return;}memset(m_pFrameBuffer,0,FRAME_BUF_SIZE);//fMaxSize =  FMAX;printf("[H264LiveVideoSource] fMaxSize:%d\n",fMaxSize);
}H264LiveVideoSource::~H264LiveVideoSource()
{envir().taskScheduler().unscheduleDelayedTask(m_pToken);if(m_pFrameBuffer){delete[] m_pFrameBuffer;m_pFrameBuffer = NULL;}if(fTruncatedBytes){delete[] fTruncatedBytes;fTruncatedBytes = NULL;}
}
int H264LiveVideoSource::maxFrameSize()
{return FRAME_BUF_SIZE;
}
void H264LiveVideoSource::doGetNextFrame()
{int uSecsToDelay = 40000; // 40 msm_pToken  = envir().taskScheduler().scheduleDelayedTask(uSecsToDelay,(TaskFunc*)getNextFrame, this);//printf("m_pToken =%p \n" ,m_pToken);
}
void H264LiveVideoSource::getNextFrame(void *ptr)
{H264LiveVideoSource *p=(H264LiveVideoSource *)ptr;if(NULL  == p)printf("null point \n");p->GetFrameData();
}#include <sys/types.h>
#include <sys/stat.h>
#include <string.h>
#include <fcntl.h>
#include <unistd.h>
#include <limits.h>typedef struct
{unsigned long long timeTick;            //时间(ms)unsigned int dataLen;                   //数据长度unsigned char dataType;                   //数据类型(DataType_E)unsigned char rsv[3]; unsigned long long timeStamp;       //编码时间戳(us)unsigned char iFrame;                //是否为关键帧unsigned char frameRate;            //帧率int encodeType;            //编码类型VideoEncodeType_Eunsigned short width;                //视频宽度unsigned short height;                //视频高度unsigned char rsv1[8];unsigned char data[0];
}IFVFrameHeader_S;
void H264LiveVideoSource::GetFrameData()
{
#if 1//memcpy(fTo,m_pFrameBuffer,fFrameSize);int read = ShareBufGetOneFrame(g_BufHandle[0], FRAME_BUF_SIZE, (char *)m_pFrameBuffer); //这里要改成你自己实际获取视频帧的函数if (read == 0){printf("read byte =0 \n");fFrameSize =0;// FramedSource::afterGetting(this);return;  }IFVFrameHeader_S *pFrameHead  = reinterpret_cast<IFVFrameHeader_S *>(m_pFrameBuffer);if(pFrameHead == NULL ){printf("pFrameHead =0 \n");fFrameSize =0;return;}if(iframetype == 0){ if(1==pFrameHead->iFrame){iframetype =1;}else{//printf("no i frame \n");//fFrameSize =0;//gettimeofday(&fPresentationTime,NULL);//FramedSource::afterGetting(this);//return;}}int framelen=pFrameHead->dataLen;#if 0if(pFrameHead->dataLen > fMaxSize)pFrameHead->dataLen = fMaxSize;memcpy(fTo,m_pFrameBuffer+sizeof(IFVFrameHeader_S),pFrameHead->dataLen);fFrameSize =pFrameHead->dataLen;#else//printf("pFrameHead->dataLen =%d fMaxSize=%u\n",pFrameHead->dataLen,fMaxSize);if(framelen > fMaxSize){framelen = fMaxSize;fNumTruncatedBytes = pFrameHead->dataLen-fMaxSize;memcpy(fTo,pFrameHead->data,framelen);memmove(fTruncatedBytes,pFrameHead->data + framelen,fNumTruncatedBytes);fFrameSize =framelen;}else{if(fNumTruncatedBytes > 0 ) {memmove(fTo,fTruncatedBytes,fTruncatedBytesNum);memmove(fTo + fTruncatedBytesNum,pFrameHead->data,framelen);fFrameSize += fTruncatedBytesNum;// printf("send last truncted %d bytes\n",fTruncatedBytesNum);fTruncatedBytesNum = 0;}else{memcpy(fTo,pFrameHead->data,framelen);fFrameSize =framelen;   }}#endiffDurationInMicroseconds = 1000000/25;gettimeofday(&fPresentationTime,NULL);//*nextPT=fPresentationTime;FramedSource::afterGetting(this);
#else#define FIFO_NAME "./test.264"//#define BUFFER_SIZE (30000)static int fd=-1;//static u_int64_t fFileSize =0;if(fd ==-1){fd = open(FIFO_NAME,O_RDONLY);if(fd > 0){// fFileSize =5316637;// GetFileSize(FIFO_NAME, fd);}}if(fd ==-1){printf("open file %s fail \n",FIFO_NAME);return;}int len =0;int remain = fMaxSize;//if(remain >fMaxSize)//    remain =fMaxSize;if((len = read(fd,fTo,remain))>0){//memmove(fTo,m_pFrameBuffer,len);gettimeofday(&fPresentationTime, NULL);fFrameSize=len;}else{if(fd >0){::close(fd);fd = -1;//printf("GetFrameData close file %d\n",len);}}fDurationInMicroseconds = 1000000/25;gettimeofday(&fPresentationTime,NULL);// printf("fMaxSize=%d fFrameSize=%d\n",fMaxSize,fFrameSize);//nextTask() = envir().taskScheduler().scheduleDelayedTask(0,//           (TaskFunc*)FramedSource::afterGetting, this);FramedSource::afterGetting(this);#endif
}

linve555常用修改点：

1, 输入的一帧数据最大值
StreamParser.cpp
#define BANK_SIZE 1500000 //帧越大，这个值就要越大

2， rtp buffer最大值
(1)Source端使用 MultiFramedRTPSource.cpp
BufferedPacket::BufferedPacket()
定义输入Buffer的上限值，即BufferedPacket的最大值
#define MAX_PACKET_SIZE 65536
(2)Sink端使用 MultiFramedRTPSink.cpp
#define RTP_PAYLOAD_MAX_SIZE 1456 //(1500-14-20-8)/4 *4 //ethernet=14,IP=20, UDP=8, a multiple of 4 bytes
MediaSink.cpp
静态变量OutPacketBuffer::maxSize = 600000; // allow for some possibly large H.265 frames，2000000 is by default
最好是RTP_PAYLOAD_MAX_SIZE的整数倍
值小了，会不断打印信息： Correct this by increasing “OutPacketBuffer::maxSize” to at least

,3,获取IP地址失败
RTSPServer::rtspURLPrefix(){
ourIPAddress(envir())
}

GroupsockHelper.cpp

 ourIPAddress(){if (badAddressForUs(from)) {#if 0char tmp[100];sprintf(tmp, "This computer has an invalid IP address: %s", AddressString(from).val());env.setResultMsg(tmp);from = 0;#endifstruct ifreq req;int ret = 0;char szIpBuf[32];sock = socket(AF_INET, SOCK_DGRAM, 0);if (-1 != sock){memset(&req, 0, sizeof(req));strncpy(req.ifr_name, "eth0", sizeof(req.ifr_name));ret = ioctl(sock, SIOCGIFADDR, &req);if (-1 == ret){close(sock);}else{memset(&szIpBuf, 0, sizeof(szIpBuf));strcpy(szIpBuf, inet_ntoa(((struct sockaddr_in *)&req.ifr_addr)->sin_addr));close(sock);fromAddr.sin_addr.s_addr=our_inet_addr(szIpBuf);from = fromAddr.sin_addr.s_addr;}}else{char tmp[100];sprintf(tmp, "This computer has an invalid IP address: %s", AddressString(from).val());env.setResultMsg(tmp);from = 0;}}

3，内存泄漏点
RTCPInstance::processIncomingReport
if(NULL != reason)
{
delete[] reason;
reason = NULL;
}
在申请内存时加上上面释放语句
reason = new char[reasonLength + 1];
4，fill sei data DeltaTfiDivisor
H264or5VideoStreamParser::H264or5VideoStreamParser()
{
//according to H264 and H265 spec, if not fill sei data, then frame_field_info_present_flag is zero. so need to set DeltaTfiDivisor to 2.0 in H264 and 1.0 in H265
if(fHNumber == 264) {
DeltaTfiDivisor = 2.0;
} else {
DeltaTfiDivisor = 1.0;
}

}
5,长时间拉取拉取RTSP流
报错误"Hit limit when reading incoming packet over TCP"
可考虑提高maxRTCPPacketSize的值
RTCP.CPP
static unsigned const maxRTCPPacketSize = 1456;

6，如播放越久延时越大
MultiFramedRTPSink.cpp->MultiFramedRTPSink::sendPacketIfNecessary() 最后延时列队uSecondsToGo 每帧都有延时时间。将uSecondsToGo 值赋为0。

7, 裁剪

只需留下这些目录，其它可删除掉。
其中liveMedia目录下有很多类型的文件，不需要的也可删除，同时修改
MediaSubsession::createSourceObjects（）把相关类型的createNew也删除掉，否则编译失败。

Live555实时视频流应用总结相关推荐

p2p传输实时视频流
HYP2P是好游科技开发的p2p sdk,利用nat穿透的原理实现p2p打洞,主要用于实时音视频的传输.使用时可以配置成tcp模式或udp模式,tcp模式会自动帮您解决丢包.乱序.重传问题.中转模式则 ...
使用Python，OpenCV实现图像和实时视频流中的人脸模糊和马赛克
使用Python,OpenCV实现图像和实时视频流中的人脸模糊和人脸马赛克 1. 效果图 2. 原理 2.1 什么是人脸模糊,如何将其用于人脸匿名化? 2.2 执行人脸模糊/匿名化的步骤 3. 源码 ...
基于 OpenCV 的网络实时视频流传输
作者 | 努比来源 | 小白学视觉大多数人会选择使用IP摄像机(Internet协议摄像机)而不是CCTV(闭路电视),因为它们具有更高的分辨率并降低了布线成本.在本文中,我们将重点介绍IP摄像机 ...
基于OpenCV的网络实时视频流传输
点击上方"小白学视觉",选择加"星标"或"置顶" 重磅干货,第一时间送达很多小伙伴都不会在家里或者办公室安装网络摄像头或监视摄像头.但是有 ...
python编程实例视屏-python实现实时视频流播放代码实例
这篇文章主要介绍了python实现实时视频流播放代码实例,文中通过示例代码介绍的非常详细,对大家的学习或者工作具有一定的参考学习价值,需要的朋友可以参考下 @action(methods=['GET' ...
python实现流媒体传输_基于OpenCV的网络实时视频流传输的实现
很多小伙伴都不会在家里或者办公室安装网络摄像头或监视摄像头.但是有时,大家又希望能够随时随地观看视频直播. 大多数人会选择使用IP摄像机(Internet协议摄像机)而不是CCTV(闭路电视),因为它 ...
利用flask将opencv实时视频流输出到浏览器
opencv通过webcam可以获取本地实时视频流,但是如果需要将视频流共享给其他机器调用,就可以将利用flask框架构建一个实时视频流服务器,然后其他机器可以通过向这个服务器发送请求来获取这台机器上 ...
H5解码H264实时视频流
浏览器如何解码实时视频流?最近研究了一下,大体思路为通过websocket把裸H264传输到浏览器,在通过js封装成mp4格式,再通过Html5的video标签进行解码,效果还是比较不错. <! ...
新的Google Lyra音频编解码器对实时视频流意味着什么？
正文字数:2602 阅读时长:4分钟通过语言编码中的码率缩减趋势,Lyra与Opus中的区别比较,Lyra的作用,XDN平台上的高效语音编码技术几个方面探讨新的Google Lyra音频编解码器对 ...
python播放视频代码_python实现实时视频流播放代码实例
这篇文章主要介绍了python实现实时视频流播放代码实例,文中通过示例代码介绍的非常详细,对大家的学习或者工作具有一定的参考学习价值,需要的朋友可以参考下 @action(methods=['GET' ...

Live555实时视频流应用总结

Live555实时视频流应用总结相关推荐

最新文章

热门文章