Honey,I Shrunk the VIDEO!

Video compression is necessary for efficient transmission and storage of video feeds from edge devices to back-end recording devices and servers. As adoption of hybrid and HD systems accelerates, manufacturers and solution providers must now address video compression and reusability issues in order to better help the integrator/installer community and expand market share.

With the transition to HD picking up pace, exciting applications in video surveillance are enabled by the unprecedented clarity that today's image sensors and cameras can provide. However, these newfound wonders require significant overhead costs in terms of bandwidth and storage capacity. The goal of video compression should be minimizing transmission and storage requirements while maintaining usable, evidentiary-grade image quality. Three thingsareats take:the source,the pipe and the destination.

 

Mark Oliver

The resolution of the image dictates its usefulness whether in terms of coverage of an area or in terms of pixels per foot for video content analysis (VCA), said Mark Oliver, Director of Product Marketing at Stretch. “Higherresolution images can be used to increase field of view or provide digital PTZ enhancements. Similarly, they can put more pixels into an area of a scene, giving clearer images and better VCA results.”

However, higher-resolution images contain more information and require more bandwidth in the transport portion of the application. This dictates the compression scheme to be used and the level of complexity to be incurred at both the compression and decompression ends of the 'pipe,' Oliver continued.                                                                                                                        
                                                                                                                                                                                

The destination of the video, the end application, must also be taken into consideration. “Careful selection of resolution and compression scheme is required if video is to be cost-effectively stored, decompressed using the potentially limited computing capabilities of the client device, or just to provide video with acceptable latency,” Oliver warned.

Two major compression standard susedtodayare M-JPEG and H.264, and there are fundamental differences. M-JPEG is an intraframe-based compression scheme where each frame is an independent image, compressed one at a time with no dependency across the frames. The advantage is that each frame can be decoded without depending on the surrounding frames and can be easily indexed down to the frame level, said Jeremiah Golston, TI Fellow and Video CTO at Texas Instruments.

However, this format is less efficient than modern compression schemes such as H.264, which use interframe coding to remove redundancy from one image to the next. H.264 gives a much higher compression ratio, and that is a big driver in the market, Golston said.

In most H.264 implementations, an intraframe is encoded as an individual image at an interval specified by the vendor, and the number of I frames used each second greatly affects compression efficiency. For a one-second, 30-fps video clip, M-JPEG encodes all 30 frames individually. “With H.264, you have almost infinite flexibility in how you use the coding techniques. You can encode every frame as an I frame, just as M-JPEG does, but still have a higher compression ratio higher than using M-JPEG,” Golston said.

Many of the reasons that M-JPEG is popular, such as interoperability and cost, have diminished over time. In the past, most browsers or mobile devices supported JPEG while H.264 had little support; now, H.264 support is almost ubiquitous. The processing horsepower demanded by H.264 for encoding and decoding is also less of an obstacle now.

 

Bach Chen

One reason M-JPEG is still used is for evidentiary purposes. “In most countries, only intracoded images can be presented as effectual evidence. That is why most network cameras still provide the M-JPEG option in addition to H.264, which is often not accepted by courts and insurers due to the motion estimation that makes it so efficient in encoding moving pictures,” said Bach Chen, Director of Platform Technology at Huper Laboratories. “However, from a technical standpoint, H.264 can provide everything M-JPEG can.” In theory, H.264 can already embed five frames per second that are encoded as stand-alone pictures and serve that same purpose, but courts want frames that are encoded as frames and do not want prediction in surrounding frames, Golston added. “Some DVRs encode both formats at the same time, using H.264 for motion and a few JPEGs per second for evidentiary purposes. .”