Analyze video surveillance big data applications and information mining

The big data application in the field is mainly reflected in two aspects: the cluster of video recording And video structured data query and information mining.

1. Cluster storage of video recording

In the big data-oriented architecture, one or more clusters can be set up according to the actual site deployment requirements. The collected stream data is divided into segments and distributed in the data cluster nodes because the cluster nodes have multiple copies of the internal backup. Such mechanisms can be ensured by software technology to ensure high reliability and stability of the overall system. These data nodes can use inexpensive general-purpose hardware, avoiding the traditional high-end hardware model, which can greatly reduce the investment cost.

Cluster storage of video files, domestic cloud storage manufacturers use CEPH technology and HDFS technology. The HDFS is used as an example. The idea is to upload the received video stream file from the local to the HDFS through the API structure provided by HADOOP. In this process, the received video file is continuously stored in a specified local temporary folder, and the local folder is dynamically changed, and the folder can be regarded as a "buffer". The files in the "buffer" will be streamed to HDFS.

解析视频监控大数据应用及信息挖掘

2. Video structured data query and information mining

The original video image is an unstructured data, which cannot be directly read and recognized by the computer and the upper application software. In order to make the video image better applied, the video image must be structured and extracted to extract key information. And perform a semantic description of the text, which is video structuring.

In a video, there are two main types of key information to be extracted: the first category is the recognition of moving targets, that is, the recognition of moving objects in the picture, whether it is a human or a motor vehicle or a non-motor vehicle; the second type is the characteristics of moving targets. Identification, that is, the characteristics of people, cars and objects in the picture, pedestrian characteristics mainly include: whether to wear glasses, scarves, tops, pants, whether to wear masks, whether backpacks, gender classification, etc.; the main characteristics of motor vehicles are: license plates Number, body color, model, etc.; object features are: size, color, direction, etc.

A case review requires a more extensive view of the relevant camera video, and the amount of video reviewed is often hundreds of thousands of hours. Video structured extraction technology extracts moving objects in the video, and then retrieves and excludes them through software, which can greatly improve the efficiency of case handling.

0 times
Window._bd_share_config = { "common": { "bdSnsKey": {}, "bdText": "", "bdMini": "2", "bdMiniList": false, "bdPic": "", "bdStyle": " 0", "bdSize": "24" }, "share": {}, "image": { "viewList": ["qzone", "tsina", "tqq", "renren", "weixin"], "viewText": "Share to:", "viewSize": "16" }, "selectShare": { "bdContainerClass": null, "bdSelectMiniList": ["qzone", "tsina", "tqq", "renren" , "weixin"] } }; with (document) 0[(getElementsByTagName('head')[0] || body).appendChild(createElement('script')).src = 'http://bdimg.share. Baidu.com/static/api/js/share.js?v=89860593.js?cdnversion=' + ~(-new Date() / 36e5)];

Plastic Road Stud

Plastic Road Stud,Led Plastic Road Stud,Wired Road Stud,Wired Aluminum Road Stud

Shenzhen Wide Way Optoelectronics Co., Ltd. , https://www.wdmtrafficlight.com