记录监控摄像头的接入过程及web端播放
Posted shame11
tags:
篇首语:本文由小常识网(cha138.com)小编为大家整理,主要介绍了记录监控摄像头的接入过程及web端播放相关的知识,希望对你有一定的参考价值。
1.rtsp视频流网页播放概述
需求:当我们通过ONVIF协议,获取到了摄像头的rtsp流地址(长这样:rtsp://admin:123456789@192.168.9.16:554/cam/realmonitor?channel=1&subtype=1&unicast=true&proto=Onvif)后,通过vlc播放器,我们可以查看监控视频内容,可是,我们应该如何在网页上查看视频内容呢?因为现在的浏览器都不支持rtsp流(详见:https://blog.csdn.net/SY__CSDN/article/details/129255690?utm_medium=distribute.pc_relevant.none-task-blog-2defaultbaidujs_baidulandingword~default-0-129255690-blog-113454774.pc_relevant_landingrelevant&spm=1001.2101.3001.4242.1&utm_relevant_index=3),因此我所选用的解决方案便是推流 + 转码
(1)转码推流工具ffmpeg(安装教程详见:https://www.cnblogs.com/h2285409/p/16982120.html),安装好之后,便可使用命令 ffmpeg -re -rtsp_transport tcp -i rtsp://admin:123456789@192.168.9.16:554/cam/realmonitor?channel=1&subtype=1&unicast=true&proto=Onvif -c:v copy -c:a copy -f flv rtmp://127.0.0.1/live/16 将我们的rtsp视频流转码并推至流媒体服务器上,在这个命令中含有两个URL,前面的是我们的rtsp流地址,而后面的URL便是我们流媒体服务器的地址,以及一个-f参数,指定了我们视频流转码后的格式为flv
(2)流媒体服务器,主要调研了2款,一是整合了Rtmp模块的Nginx,二是SRS视频服务器,而我所选用的是SRS(官方文档:http://ossrs.net/lts/zh-cn/),在使用ffmpeg推流上SRS后,便可直接从SRS获得HTTP-FLV视频流地址(如本例:http://127.0.0.1/live/16.flv ),然后,前端通过flv.js组件库便可直接在页面上播放该视频流
SRS与ffmpeg参考:https://blog.csdn.net/diyangxia/article/details/120172920
ffmpeg进阶参考:https://segmentfault.com/a/1190000039782685
flv.js参考:http://www.kaotop.com/it/446261.html
2.rtsp推流转码相关代码实现
//ffmpeg安装路径
@Value("$ffmpegPath")
private String ffmpegPathPrefix;
//srs视频服务器地址
@Value("$srsAddress")
private String srsAddress;
//srs端口,默认为8080
@Value("$srsPort")
private String srsPort;
//srs-http-api端口,默认为1985
@Value("$srsHttpApiPort")
private String srsHttpApiPort;
@Resource
private MonitorMapper monitorMapper;
@Resource
private RedisTemplate redisTemplate;
@Resource
private RestTemplate restTemplate;
@Resource
private ThreadPoolTaskExecutor threadPoolTaskExecutor;
private ConcurrentHashMap<String, TranscodeModel> id2transcodeModelMap = new ConcurrentHashMap<>();
/**
* 进行推流转码
* @param id ipc的主键id
* @return 转码推流后的http-flv地址,前端可通过flv.js直接播放
*/
public String transcodeAndPushStream(String id)
Ipc ipc = monitorMapper.getIpcInfoById(id);
try
//先给这个流加锁,防止其他用户请求该流信息
while(!redisTemplate.opsForValue().setIfAbsent(id, 1, Duration.ofSeconds(60)))
Thread.sleep(200);
//避免重复对某一个流的推流工作
if(!id2transcodeModelMap.containsKey(id))
String command = String.format("%sffmpeg -re -rtsp_transport tcp -i %s -c:v copy -c:a copy -f flv %s",this.ffmpegPathPrefix, ipc.getRtspUrl(), "rtmp://" + this.srsAddress + "/live/" + id);
//通过命令行执行推流转码
System.out.println("启动推流转码, 其命令为: " + command);
Process process = Runtime.getRuntime().exec(command);
//可选,开启异步线程,观察推流进程所打印的日志
Future<Void> processOutputHandler = threadPoolTaskExecutor.submit(() ->
BufferedReader br = new BufferedReader(new InputStreamReader(process.getErrorStream()));
String msg = null;
try
while ((msg = br.readLine()) != null)
if (Thread.currentThread().isInterrupted())
System.out.println("关闭推流进程的日志输出线程: " + Thread.currentThread().getName());
break;
if (msg.contains("fail") || msg.contains("miss"))
System.err.println(ipc.getId() + " 在推流过程中发生故障或丢包: " + msg);
catch (IOException e)
System.err.println(ipc.getId() + " 在推流转码过程中发生异常错误,原因: " + e.getMessage());
finally
if (Thread.currentThread().isAlive())
Thread.currentThread().interrupt();
return null;
);
id2transcodeModelMap.put(id, new TranscodeModel(id, command, process, processOutputHandler));
catch (Exception e)
this.closeTranscode(id);
throw new RuntimeException("启动对 " + id + " 的推流转码失败,原因: " + e.getMessage());
finally
redisTemplate.delete(id);
//返回转码后的flv流地址
return "http://" + this.srsAddress + ":" + this.srsPort + "/live/" + id + ".flv";
/**
* 关闭推流进程
*/
private void closeTranscode(String id)
TranscodeModel transcodeModel = null;
if((transcodeModel = id2transcodeModelMap.get(id)) != null)
Future<Void> outputHandler = transcodeModel.getOutputHandler();
//关闭输出线程
if(outputHandler != null && !outputHandler.isDone())
outputHandler.cancel(true);
//停止推流转码进程
if (transcodeModel.getProcess() != null)
transcodeModel.getProcess().destroy();
id2transcodeModelMap.remove(id);
System.out.println("关闭对 " + id + " 的推流转码");
/**
* 客户端结束播放流后,srs可配置触发一个on_stop回调,通过该回调,我们就可以知道哪些流可能没人看了,然后结束对该流进行的推流转码工作
* @param data srs触发回调时所携带的参数
*/
public void stopPlay(CallbackOnStopPlay data)
String clientId = data.getClient_id();
JSONObject srsClient = this.requestSrsClientById(clientId);
if(!srsClient.isEmpty())
String streamId = srsClient.getString("stream");
if (!StringUtils.hasText(streamId))
System.err.println("获取client " + clientId +" 的流失败, 未关联流");
return;
//在请求这个流的信息之前,先给这个流加锁,防止其他用户预览该流
try
while(!redisTemplate.opsForValue().setIfAbsent(data.getStream(), 1, Duration.ofSeconds(60)))
Thread.sleep(200);
JSONObject vidiconStream = this.requestSrsStreamById(streamId);
if(!vidiconStream.isEmpty())
Integer clients = vidiconStream.getInteger("clients");
//当前观看该流的人数 <= 2时,说明没人看了可以停止推流,至于为什么是2,可以自己观察打印日志看看
if(clients <= 2)
this.closeTranscode(vidiconStream.getString("name"));
catch (Exception e)
System.err.println("关闭视频流 " + streamId + " 失败, 原因: " + e.getMessage());
finally
redisTemplate.delete(data.getStream());
/**
* 根据clientId获取某个client信息
*/
private JSONObject requestSrsClientById(String clientId)
if(!StringUtils.hasText(clientId))
return new JSONObject();
String url = "http://" + this.srsAddress + ":" + this.srsHttpApiPort + "/api/v1/clients/" + clientId;
ResponseEntity<JSONObject> exchange = null;
try
exchange = restTemplate.exchange(url, HttpMethod.GET, new HttpEntity<>(null, new HttpHeaders()), JSONObject.class);
catch (Exception e)
System.err.println("请求srs的client " + clientId + " 失败,原因: " + e.getMessage());
return new JSONObject();
if (exchange == null || exchange.getBody() == null || exchange.getBody().getInteger("code") != 0)
System.err.println("请求srs中client " + clientId + " 失败");
return new JSONObject();
System.out.println("请求到client " + clientId + " 的信息为: " + exchange.getBody());
return exchange.getBody().getJSONObject("client");
/**
* 根据流的id获取某个流
*/
private JSONObject requestSrsStreamById(String streamId)
if(!StringUtils.hasText(streamId))
return new JSONObject();
String url = "http://" + this.srsAddress + ":" + this.srsHttpApiPort + "/api/v1/streams/" + streamId;
ResponseEntity<JSONObject> exchange = null;
try
exchange = restTemplate.exchange(url, HttpMethod.GET, new HttpEntity<>(null, new HttpHeaders()), JSONObject.class);
catch (Exception e)
System.err.println("请求srs中的流 " + streamId + " 失败,原因: " + e.getMessage());
return new JSONObject();
if (exchange == null || exchange.getBody() == null || exchange.getBody().getInteger("code") != 0)
System.err.println("请求srs中的流 " + streamId + " 失败, 由于服务器重启或其他原因,该流已失效");
return new JSONObject();
System.out.println("请求到流 " + streamId + " 的信息为: " + exchange.getBody());
return exchange.getBody().getJSONObject("stream");
public class TranscodeModel
private String id;
private String command;
private Process process;
//推流过程中的输出线程
private Future<Void> outputHandler;
//客户端关闭流时触发的回调所传递的参数
public class CallbackOnStopPlay
private String server_id;
private String action;
private String client_id;
private String ip;
private String vhost;
private String app;
private String stream;
private String param;
//ipc类
public class Ipc
//ipc的主键id
private String id;
//ipc的rtsp流地址
private String rtspUrl;
对推流进程的关闭,可以选择定时任务轮询srs中流的信息,然后对那些没人看的流进行关闭,也可以选择配置srs客户端关闭流时的回调,来进行关闭,至于回调如何配置使用,可以详看官方文档中开放接口相关内容和这篇文章:https://blog.csdn.net/weixin_44341110/article/details/120829847
3.通过海康,大华NVR来接入IPC
未完待续...
以上是关于记录监控摄像头的接入过程及web端播放的主要内容,如果未能解决你的问题,请参考以下文章
监控摄像头GB28181转RTMPFLVHLSRTSP多种格式实现网页嵌入播放