如何解析DASH流MPD文件获取分片与分辨率参数用于视频渲染及码率切换

基础解析规则 #

你要的分辨率、分片信息都可以直接从MPD的XML标签属性里提取，核心规则如下：

• 分辨率、流标识、码率：从<Representation>标签读取，id是流唯一标识，width/height对应视频宽高，bandwidth是码率（单位bps），codecs是编码格式。
• 分片信息：你给出的样例用的是SegmentTemplate模板分片方案，读取以下属性即可：
  • initialization：流初始化分片地址，播放任意码率的流都需要先请求该文件
  • media：媒体分片模板，其中$Number$是分片序号占位符
  • startNumber：分片起始序号，模板里的$Number$从该值开始递增
  • 单分片时长 = duration / timescale，你的样例里单分片时长为 25008 / 24000 = 1.042秒

你提供的MPD片段解析 #

原始MPD片段 #

<SupplementalProperty schemeIdUri="urn:mpeg:dash:srd:2014" value="0,0,0,1,1,3,3"/>
<Representation id="1" mimeType="video/mp4" codecs="hev2.1.6.L186.0" width="640" height="360" frameRate="24" sar="1:1" startWithSAP="1" bandwidth="131694">
 <SegmentTemplate timescale="24000" media="sequence_qp22_dash_track1_$Number$.m4s" startNumber="1" duration="25008" initialization="hevc_srd_set1_init.mp4"/>
</Representation>
</AdaptationSet>
<AdaptationSet segmentAlignment="true" maxWidth="640" maxHeight="360" maxFrameRate="24" par="16:9" lang="und">
<SupplementalProperty schemeIdUri="urn:mpeg:dash:srd:2014" value="0,1,0,1,1,3,3"/>
<Representation id="3" mimeType="video/mp4" codecs="avc1.4d401e" width="640" height="360" frameRate="24" sar="1:1" startWithSAP="1" bandwidth="131424">
 <SegmentTemplate timescale="24000" media="tile2-360p-128kbps_dash$Number$.m4s" startNumber="1" duration="25008" initialization="tile2-360p-128kbps_dashinit.mp4"/>
</Representation>
<Representation id="4" mimeType="video/mp4" codecs="avc1.4d401e" width="640" height="360" frameRate="24" sar="1:1" startWithSAP="1" bandwidth="1492186">
 <SegmentTemplate timescale="24000" media="tile2-360p-1500kbps_dash$Number$.m4s" startNumber="1" duration="25008" initialization="tile2-360p-1500kbps_dashinit.mp4"/>
</Representation>
</AdaptationSet>
<AdaptationSet segmentAlignment="true" maxWidth="640" maxHeight="360" maxFrameRate="24" par="16:9" lang="und">
<SupplementalProperty schemeIdUri="urn:mpeg:dash:srd:2014" value="0,2,0,1,1,3,3"/>
<Representation id="5" mimeType="video/mp4" codecs="avc1.4d401e" width="640" height="360" frameRate="24" sar="1:1" startWithSAP="1" bandwidth="129740">
 <SegmentTemplate timescale="24000" media="tile3-360p-128kbps_dash$Number$.m4s" startNumber="1" duration="25008" initialization="tile3-360p-128kbps_dashinit.mp4"/>
</Representation>
<Representation id="6" mimeType="video/mp4" codecs="avc1.4d401e" width="640" height="360" frameRate="24" sar="1:1" startWithSAP="1" bandwidth="1451324">
 <SegmentTemplate timescale="24000" media="tile3-360p-1500kbps_dash$Number$.m4s" startNumber="1" duration="25008" initialization="tile3-360p-1500kbps_dashinit.mp4"/>
</Representation>
</AdaptationSet>

解析后参数数组示例 #

你可以直接按如下结构整理参数，用于后续渲染或码率切换：

const dashStreamList = [
  {
    id: "1",
    codec: "hev2.1.6.L186.0",
    resolution: "640x360",
    bandwidth: 131694,
    initSegmentUrl: "hevc_srd_set1_init.mp4",
    segmentTemplate: "sequence_qp22_dash_track1_$Number$.m4s",
    startSegmentNo: 1,
    segmentDuration: 1.042,
    srdTileX: 0 // 从SupplementalProperty的value第2位获取，用于拼接渲染
  },
  {
    id: "3",
    codec: "avc1.4d401e",
    resolution: "640x360",
    bandwidth: 131424,
    initSegmentUrl: "tile2-360p-128kbps_dashinit.mp4",
    segmentTemplate: "tile2-360p-128kbps_dash$Number$.m4s",
    startSegmentNo: 1,
    segmentDuration: 1.042,
    srdTileX: 1
  },
  {
    id: "4",
    codec: "avc1.4d401e",
    resolution: "640x360",
    bandwidth: 1492186,
    initSegmentUrl: "tile2-360p-1500kbps_dashinit.mp4",
    segmentTemplate: "tile2-360p-1500kbps_dash$Number$.m4s",
    startSegmentNo: 1,
    segmentDuration: 1.042,
    srdTileX: 1
  },
  {
    id: "5",
    codec: "avc1.4d401e",
    resolution: "640x360",
    bandwidth: 129740,
    initSegmentUrl: "tile3-360p-128kbps_dashinit.mp4",
    segmentTemplate: "tile3-360p-128kbps_dash$Number$.m4s",
    startSegmentNo: 1,
    segmentDuration: 1.042,
    srdTileX: 2
  },
  {
    id: "6",
    codec: "avc1.4d401e",
    resolution: "640x360",
    bandwidth: 1451324,
    initSegmentUrl: "tile3-360p-1500kbps_dashinit.mp4",
    segmentTemplate: "tile3-360p-1500kbps_dash$Number$.m4s",
    startSegmentNo: 1,
    segmentDuration: 1.042,
    srdTileX: 2
  }
]

额外说明 #

你提供的MPD片段里包含urn:mpeg:dash:srd:2014的补充属性，这是DASH的空间区域描述标识，value字段的第2位是tile的X坐标，第3位是Y坐标，你的样例是3列1行的拼接流，渲染时需要把三个tile按X坐标从左到右拼接成完整画面。

功能实现方法 #

• 分辨率切换：根据当前网络带宽匹配对应bandwidth的流，切换时先加载新流的初始化分片，再按当前播放进度计算对应分片序号，请求新流的对应媒体分片即可，无需从头播放。
• 生成分片地址：将segmentTemplate里的$Number$替换为从startSegmentNo开始的递增整数，即可得到所有分片的完整地址。

内容的提问来源于stack exchange，提问作者穿拖鞋的小涵