Because video surveillance footage can hold the key to solving a crime and degradation can undermine the quality of the video, widespread camera usage has resulted in an increased need for video analysis and enhancement by technical investigators – enabling them to unlock critical details that might previously have been lost without analysis tools. Fortunately for investigators, advances in the processing power of personal computers now give them the ability to use simple computer-based digital video capture and editing systems to enhance and clarify video content.
Before engaging in video clarification, a technical investigator should understand the most common sources of video degradation. Degradation can be caused by environmental sources, video camera quality and type, the transmission of the video signal from a camera to the recorder, the video recorder quality and the actual tape from which the video signal was recorded. These sources create various effects that impede the technical investigator’s ability to create an accurate, detailed image of the recorded scene.
The most common causes of video degradation, outlined in Figure 1, are atmospheric effects, noise, motion blur, out-of-focus blur, low image resolution, low signal bandwidth and tape and wear stretch. With some planning, careful installation of equipment and the implementation of video enhancement tools now specially designed for law enforcement and security personnel, these common pitfalls can become a thing of the past.
Atmospheric effectsEven before light enters the camera, atmospheric perturbations can cause light to bend and shift. Common examples of this are mirage effects from heated air and visibility reduction from smog, cloud cover or rain. Poor lighting is another common cause of video problems. Adequate, balanced lighting can improve the quality of video captured, and backlit; track lighting can cast shadows that degrade video quality. See Figure 2.
There are a variety of simple filtering techniques available on PC-based editing systems that allow security personnel to easily enhance video tainted by atmospheric effects, with amazing accuracy. In one example, video taken of a vehicle traveling in a rainstorm was enhanced by using an image stabilization filter – to stabilize the effects of motion from the in-car camera – frame averaging – which more clearly assembles pieces of the image based on the average of multiple frames in a series – and sharpening. The make of the vehicle and the license plate can clearly be seen after these techniques are used. What’s more, this process takes only minutes to execute using appropriate software designed specifically for law enforcement and security officers.
NoiseNoise is an unwanted signal that corrupts an original image. The most common form of noise is random white noise, such as the so-called “snow” seen when viewing weak television broadcasts. Random white noise exists everywhere in nature and is virtually impossible to avoid. See Figure 3.
Even the heat of the circuitry in a camera generates its own white background noise called thermal noise. For low-light images, some external forces can create random variations of electric charge on video camera’s charge-coupled device (CCD), thus creating snow in the image. White noise can become particularly disruptive and more apparent when the video is amplified to the desired viewing level. So one might particularly encounter this issue when scouring video for the details, the very time that it is most important to see clearly. That is why video enhancement is so critical. Video forensics software enables security professionals to negate noise by using frame averaging to create a clearer composition.
Motion blurMotion blur occurs when an object is moving relatively fast in a scene. Video camera sensors integrate the effects of light over time in order to register an image. If the image content moves during this integration time, the resulting image will be blurred in the direction of motion. Many CCD cameras allow adjustment of the electronic shutter speed as appropriate for the given lighting, depth of field and the degree of motion, which can help prevent motion blur.
In Figure 4, a vehicle traveling over 60 mph passed the field of view of a video camera. The video capture of the vehicle totaled only three frames – a typical video camera captures 30 frames per second. In order to enhance this video, those three frames were de-interlaced. One field – there are 60 fields per second – was extracted to yield a single image of the truck: an image blurry from high-speed linear motion. Using a de-blur filter, the identity of the vehicle was more easily identified. This type of enhancement tool is simply applied and in minutes, compelling video evidence can be obtained.
Out-of-focus blurMost cameras use a lens or lenses to secure an image. If these optics are not properly adjusted for the distance to a given object in the field of view, the object will be blurred slightly in all directions.
Out-of-focus blur can be prevented by adjusting camera focus when installing a security system. To best determine what camera focus is most appropriate for a system, the focus should be tested under various lighting conditions. The iris automatically changes aperture under changing lighting, which can make video out of focus.
Low image resolutionSince the number of pixels in a digital video image is finite (e.g. 640 rows by 480 columns), the detail visible in an object of interest depends on the amount of the image area that it occupies. Unlike film, magnification of recorded video offers no increase in image detail; it simply makes the pixels larger. Zooming in on an image will make a small section appear larger and thus possibly easier to see, but no detail may be gained this way. Choosing proper camera placement and field of view can help ensure proper resolution.
Figure 5 shows the effects that image resolution has on the ability to recognize a face. To better explain the effects of resolution, it must be understood that a letter or number must be at least three pixels wide (plus one pixel for spacing) in order to be recognizable in an image. For a seven-letter license plate, a bare minimum of 27 pixels is needed to read the tag. This represents about 4 percent of the width of most video images. This same theory applies to images as depicted in this figure.
Low-signal bandwidthAfter the video camera converts each video image into an electrical signal, it must be transmitted over video communication cables, RF broadcasting equipment, and/or recorded to tape. These devices limit the bandwidth of signals that may be transmitted and recorded, thus limiting detail such as the sharpness of vertical edges.
Better bandwidth comes with better equipment. Also, using the same tape in older equipment can cause low bandwidth among other problems. To improve the occurrence of acceptable image acquisition, cameras should have a signal bandwidth of at least 7 MHz.
All too often the owners of video surveillance systems will neglect cleaning their video recorders or changing their videocassettes, causing videotape degradation due to both storage time and playback wear. The most common damage to tape is caused by friction from a VCR’s rollers, capstan and playback head. Wrinkling and tape stretch can also cause signal dropouts and unstable synchronization during playback. See Figure 6.
Tape wear, stretchVHS tapes should be used no more than 12 times and should be replaced after one year to maintain optimal video quality.
With the proper equipment setup and specially engineered video enhancement tools available, uncovering clearer, more detailed images from video surveillance is now easier than ever, taking just minutes to perform an analysis that would otherwise have been virtually impossible. With specialized and affordable video forensics systems on the market that are designed for the daily challenges security personnel face, the quality of the video captured on any security system instantly becomes more valuable and usable in criminal and other investigations.