Advanced Imaging


Advanced Imaging Magazine

Updated: January 12th, 2011 09:49 AM CDT

Putting the Squeeze on Surveillance Video

H.264 compression provides high resolution with variable bit and frame rates
A Mexican city’s metro station uses a Euresys Picolo V16 H.264 board to digitize, compress, locally store and broadcast over a network to a central station where it’s ready for observation and review.
The Picolo V16 video capture board features real-time H.264 on-board compression for 16 video channels with audio capability. Each video input simultaneously delivers a formatted and compressed video stream, both of which are independently configurable.

By Barry Hochfelder

The flexibility and programmability of H.264 shows up in another Euresys application—police cars. The video surveillance in the cars is connected to central dispatch via a 3G network line. The bandwidth is low, 300 kb/second, so you have to compress it more. Two streams of video are needed. One is high-resolution, high bit rate, high frame rate, and the other is low resolution, bit rate and frame rate.

“They store the image locally at high resolution and high bit rate, but they also have the very low stream transmitted wirelessly, so they get both the image stored [at central dispatch] and the one stored in the police car,” Damhaut explains.

According to its web site, the Stretch Scalable Video CODEC (SVC) uses a layered approach to video compression in which a base layer of minimum resolution, frame rate and quality is compressed first. Additional layers of incrementally higher frame rate, resolution, or quality are then encoded using only the information needed to enhance the previous layer to the desired degree. At the decoder, layers can be sequentially decoded, resulting in video of progressively higher frame rate, quality, or resolution. Decoding can be terminated when video with the desired characteristics is achieved.

The layered structure of an SVC stream means that truncating it to remove higher order layers reduces the frame rate and/or resolution of the encoded video. The lower order layers and base layer can still be decoded in the normal way to generate viewable video. To change the frame rate or resolution of a conventional AVC stream, the stream would need to be decoded to recover the video. The video could then be resized or re-sampled and then would have to be re-encoded. Simply truncating the stream as with SVC would result in a corrupted stream that could not be decoded.

The ability to truncate SVC streams means that they can easily be managed after they are stored to disk. A simple file operation is all that is required to remove unwanted layers, reduce the resolution or frame rate of the video, and recover valuable storage space. Similarly, an SVC stream can be truncated to reduce its bit rate in order to transit a network of limited bandwidth. SVC streams can be sent further out into the network and can be monitored by remote personnel more easily. Truncated streams with reduced resolution and bandwidth also require less compute bandwidth to decode. SVC streams can not only be sized according to the network characteristics, but also according to the decode capabilities of the decoding device.

Subscribe to our RSS Feeds