W H I T E PA P E R

Video Solutions: Understanding Video Quality

Many organizations are addressing the requirements for video solutions. While it is important to construct video applications support systems quickly to address immediate needs, network operators need to select the right system the rst time, to minimize the total life cycle cost of these systems.

Abstract Operators need to clearly understand the requirements of their organization. This includes an understanding of the video image quality. While there are many other issues that will signicantly impact network design, this paper provides information on factors that address image quality. The factors discussed include: Resolution Frames Per Second (FPS) Video Codec Packets Per Second (PPS) Bit Rate

medium. NTSC denes an aspect ratio of 4:3, which provides a theoretical maximum resolution of 720 (horizontal) by 486 (vertical) when specied in terms of non-square (10:11) pixels. In square pixels this would be 648x486, or more commonly 640x480. The resolution issue is complicated by additional common resolution standards for video encoding. Common Intermediate Format (CIF) is 352x288 @ 30fps. It represents a compromise between the NTSC frame rate (30fps) and PAL resolution. In practice, producing 352x288 from a NTSC camera source is difcult. As such North American surveillance vendors have created a modied CIF format of 352x240 (NTSC CIF) which is easily derived from an NTSC video source by removing every other vertical scan line. Alternatives are Quarter CIF (QCIF) 176x144 (or 176x120), 4CIF 704x576 (or 704x480), and 16CIF 1408x1152 (or 1408x960). The interaction between encoded resolution and display resolution of the decoded video can be dramatic. By understanding the device which will be used to view the decoded video, the network operator can know the requirement for image collection and processing. The resolution of the display device has an impact on overall perceived video quality, since the encoded video will need to be scaled accordingly. Frames Per Second (FPS) Frames Per second (FPS) is the number of snapshots of the video scene in one second. Recall that modern lms are 24 FPS, NTSC is 30 FPS, and PAL is 25 FPS. Video surveillance cameras can be congured for a range of FPS. In many cases, 10 FPS is sufcient. Video Codecs One might think that stipulating the encoders resolution and Frames Per Second (FPS) would exactly result in a bandwidth throughput need. However, even when stipulating both the resolution and FPS, the IP camera may be congured for a particular bandwidth within some bandwidth

An understanding of the effect of these factors on bandwidth requirements will enable network operators to correctly design communications infrastructure networks that will support video applications. Intricacies of Video Solution Image Quality Operators need to select video performance that is appropriate to the needs of the application. Different performance levels will require different cameras to collect images. In general, higher performance levels will require a higher bandwidth infrastructure to transport data from the camera to the video command center. Network operators need to clearly understand the camera performance required in order to design the communications infrastructure appropriately. If there is not sufcient capacity at any point in the communications infrastructure, video images may be delayed or lost, defeating the purpose of a video solution. Resolution Resolution is the number of pixels (picture elements) contained within each frame of video. For example, The National Television System Committee (NTSC) as a standard species a mandatory 525 analog scan lines of vertical resolution, of which 486 are typically visible. Horizontal resolution is variable dependent on the recording or display

2 WHITE PAPER - Video Solutions: Understanding Video Image Quality

range. There are tradeoffs between congurations. In general, higher bandwidth allocation for a given fps and resolution usually will lead to better video quality. An encoder device collects and produces a compressed video stream from the camera. The most common codecs used in video surveillance are MJPEG, MPEG4-SP, and H.264. (See Table 1) Table 1: Common Surveillance Video Codecs Codec Type MJPEG Codec Characteristics Independantly decodable frames Greatest Bandwidth required for a given video quality Frames are encoded in a dependant manner Designed for motion video Most common in todays surveillance industry Builds upon MPEG-4 to provide best quality vs. bandwidth required tradeoff

Data Bit Rate Packet size and transfer rate (Packets Per Second PPS) is signicant for video applications. The overhead of encapsulating a video stream into a packet stream can be substantial depending on the data requirements and network conguration. In general, encoded video will be encapsulated in Real-time Transport Protocol (RTP), RTP encapsulated in User Datagram Protocol (UDP), and UDP encapsulated in Internet Protocol version 4 (IPv4). Additionally IP is encapsulated into some Data Link Layer protocol, such as Ethernet. The overhead sum of Ethernet/IP/UDP/RTP is 18+20+8+12 = 58 octets, plus the video payload. The bit rate consumption in kbps is typically stated for a codec and associated parameters (like fps and resolution). Network Bandwidth It is intuitive that bandwidth will be predominantly owing in an uplink direction from the video encoder to the network. The network operator will need to allocate some bandwidth in the downlink direction from the network to the camera. This downlink bandwidth is used to control the camera for Pan, Tilt, Zoom (PTZ) and other functions. In many networks, the network operator can designate the up/down ratio of data to the

MPEG-4 Simple Prole (SP)

MPEG-4 Part 10, H.264, AVC

3 WHITE PAPER - Video Solutions: Understanding Video Image Quality

Table 2: Typical Quality Settings and Network Bandwidth Consumption (MPEG4-SP) Frames per Seconds Resolution QCIF CIF 4CIF 5 FPS 25 kbps 100 kbps 400 kbps 10 FPS 50 kbps 200 kbps 800 kbps 15 FPS 75 kbps 300 kbps 1200 kbps 25 FPS 125 kbps 500 kbps 2000 kbps 30 FPS 150 kbps 600 kbps 2400 kbps

cameras in terms of a percentage. For many video surveillance applications, it is common to have 10% downstream (control signals to the camera) and 90% upstream (video images from the camera). Putting It All Together While there are many factors affecting bandwidth consumption of a video stream, table 2 depicts typical quality settings and their use of network bandwidth for a variety of applications. In each case, an MPEG4-SP codec is used.

Motorola, Inc. 1301 E. Algonquin Road, Schaumburg, Illinois 60196 U.S.A. www.motorola.com/motowi4
MOTOROLA and the stylized M Logo are registered in the U.S. Patent and Trademark Ofce. All other products or service names are the property of their registered owners. Motorola, Inc. 2008