Talking about Video Image Processing Technology in Security Monitoring

With the rapid development and popularization of computer software and hardware technology, human beings have entered a fast-developing information age. About 80% of human information comes from images. Image processing technology is becoming more and more indispensable in scientific research and technology applications. s method. The security industry has entered a new era. The combination of HD, intelligence and big data and cloud technology has become a new topic of the times. Along with this, video image processing technology has gradually matured, and only in this way can it meet the development of industry technology. For this reason, research on video image processing technology has become an area worthy of attention in the security industry.

Application value of video image processing technology

Video image processing technology is simply a computer processing of digital video images, which is essentially a signal processing process and is discrete signal processing. Why do image processing in the video surveillance industry? So what do we need to analyze the advantages of video image processing?

First, improve the visual quality of the image, such as brightness, color conversion, enhancement, suppression of certain components, geometric transformation of the image, etc., to improve the quality of the video image.

Second, extracting certain features or special information contained in the video image, these extracted features or information often facilitate computer analysis of the image. The process of extracting features or information is a pre-processing of computer or computer vision. The extracted features may include many aspects such as frequency domain features, grayscale or color features, boundary features, region features, texture features, shape features, topological features, and relationship structures.

Third, video image enhancement and restoration. The purpose of image enhancement is to transform the image into a form that is more suitable for human and machine analysis. Commonly used enhancement methods are: gray level histogram processing; interference resistance; edge sharpening; pseudo color processing. The purpose of image restoration is the same as image enhancement. The main principle is to eliminate or reduce image damage and degradation caused by image acquisition and transmission, including image blur, image interference and noise, etc. Real image. Regardless of image enhancement or image restoration, all pixels of the entire image must be operated. For the large number of image pixels, the calculation is also very large. The purpose of encoding is to compress the amount of information of an image without changing the quality of the image to meet the requirements of transmission and storage. The coding mostly uses digital coding technology to process the image point by point. This is often encountered in public security for video images to restore details.

Fourth, the analysis of the image. Extract some useful metrics, data, and information from the image to a numerical result. Image analysis The image segmentation method extracts the features of the image and then symbolizes the image. This description not only answers the image whether a specific object exists, but also describes the image content in detail.

Fifth, it can make image reproducibility. The fundamental difference between digital image processing and analog image processing is that it does not degrade image quality due to a series of transformation operations such as image storage, transmission or copying. As long as the image accurately represents the original when digitized, digital image processing The process always maintains the reproduction of the image.

Sixth, flexible and variable processing functions. For the simulation technology, our current video surveillance image processing technology can divide the monitoring image into two-dimensional arrays of arbitrary size, which means that the digital precision of the image can meet the needs of any application, and it conforms to the high-definition trend of the industry.

The seventh is applicable to the width of the face. The video surveillance image can come from a variety of information sources, such as a bank screen, or a remote monitoring such as an airport or maritime. The image of the source of information from any industry can be converted into a digitally encoded form. A combination of grayscale images represented by dimensional arrays,

Eight of them, high flexibility. Digital image processing can not only perform linear operations, but also achieve nonlinear processing, that is, all operations that can be expressed by mathematical formulas or logical relationships can be realized by digital image processing.

However, the video image processing technology occupies a wide frequency band, so in the implementation of imaging, transmission, storage, processing, display and other aspects, the technical difficulty and cost are high, which requires us to understand the video image processing technology. the elements of.

Mainstream technology of video image processing technology

Video image processing involves a wide range of knowledge in terms of technology. There are many kinds of specific methods and applications are very common. However, if analyzed in the security industry, the main content of the technology includes three parts: image compression; enhancement and restoration; matching. , description and identification. Due to the requirements of ultra high definition, H.265 has a natural advantage in image compression. Since the technical topic on H.265 has already explained the relationship with video image processing, this article does not specifically introduce this aspect. To this end, in the remaining two parts, we are currently focusing on video image processing technology in the following aspects:

Intelligent analytical processing technology

Intelligent analysis and processing technology is currently the image processing technology that the Chinese security industry pays more attention to. The technology is simply to find the moving objects in the image, track and analyze them, discover abnormal behaviors in time, trigger alarms and take other measures. Intervene. The intelligent analysis processing technology is developed based on the monitoring image processing software package, and mainly has the following image processing forms:

1. Separate the (moving) target from the video image. 2. Perform behavior analysis on the target, determine the direction and mode of the movement, and discover and alarm abnormal behaviors; generate the target's motion trajectory and automatically track the target. 3. Realize the analysis of video semantics and the high level of image processing analysis. By analyzing an image sequence and obtaining the real information it contains, it can be combined with the semantic analysis of speech to gradually realize the analysis of video semantics. 4. Separate target, behavior analysis and motion tracking in complex environments, especially to achieve multi-target tracking.

The above points have been practically applied (for example, single-line intrusion detection, multi-turn line, fence intrusion, entry/exit area detection, flaw detection, residue detection, object movement detection, object appearance detection, object disappearance detection, population density) , crowd mutation, running detection, reverse detection, human flow, flame detection, smoke detection, scene change, single ball machine PTZ automatic tracking, etc., but basically in a simple environment, intelligent image processing for a few targets. Realizing these functions in a complex environment video surveillance environment (multi-person flow, multiple mobile, bad weather) is the true value of image intelligence analysis and processing technology. To solve the comprehensive analysis of multiple images, the association of targets between images, and the continuity of target tracking, this is an urgent need of market applications, and it is also an urgent problem we need to solve.

Wide dynamic technology

Wide dynamic technology is Panasonic's first attempt in the surveillance industry, which was 40 times its wide dynamic range in 1977, and was introduced to the industry in Shenzhen by the industry in 2005. Two years later, the US PIXIM program is also widely popular in the market, becoming another option for wide dynamics.

Wide dynamic video image processing technology is mainly used to solve the phenomenon that the image captured by the camera in a wide dynamic scene is overexposed and the dark area is not exposed enough. In short, the wide dynamic technique allows for particularly bright and particularly dark areas of the scene to be seen simultaneously in the final image. This kind of technology is a common video image processing technology in the security industry. After more than ten years of market baptism and technology improvement, there are still some shortcomings. First, it is due to its short exposure feature in the highlight area, which will cause horizontal stripe flicker in the bright area under the fluorescent lamp scene. This is the common flicker phenomenon caused by the power frequency of the fluorescent lamp, because the exposure time in the bright area is very short and does not reach. The minimum exposure time required for anti-flash; the second is due to its long-exposure feature in dark areas, which causes smear due to excessive exposure time when the object is moving at high speed.

Fog treatment

This technology began to be applied domestically in 2010 with the advent of HD. At that time, the movement of Hitachi launched the main function of fog. The technology can process video images in various smog weather in real time, remove the scene caused by smog, blur, invisible or invisible, and can significantly enhance the detailed information of the video image, so that the original is obscured. The image details are fully displayed and the original color tones are maintained, making the video image extremely bright, transparent and clear, thus achieving good image quality and visual experience.

Lossless real-time amplification

The video image can be zoomed in various magnifications in real time, the enlarged video image without any blur and mosaic, and the details of the original video image can be fully retained.

Strong light suppression processing

The highlights or areas of the strong light in the video image can be suppressed in real time, so that the range and brightness of the strong light are reduced, and the contour and color of the target scene hidden by the strong light scattering are restored, and a clear video image is obtained.

Remove blurring

It can process defocus blur and motion blur in the image to make the image of the observed scene clear.

The first three can be said to be the three video image processing technologies currently concerned by the security industry. The image processing of the latter three is mostly based on several mainstream product solutions in the market, so that there is not much good or bad in product performance. the difference.

Video image processing technology is special in security requirements

Because the processing speed of the computer is extremely fast, and the digital signal has the characteristics of small distortion, easy storage, easy transmission, strong anti-interference ability, computer image processing is widely used, including aerospace, telemetry, medical equipment, industrial automation detection. , security identification, security monitoring, entertainment media and other major areas. Each application field has its own specific requirements in the field, and the security monitoring industry application also has its inherent particularity.

There is a high demand for image sharpness. At the public security monitoring site, public security organs often need to monitor suspects and evidence through surveillance video. Video that is generally not sharp enough does not meet this requirement. At the traffic monitoring site, traffic police need to monitor the images to identify license plates, violations, drivers and other requirements, and blurred images cannot be applied in this case.

Different industry monitoring, the difference in image requirements. For example, medical monitoring requires a higher color reproduction of images. Intelligent traffic monitoring requires high night illumination and capture speed of the camera, and requires clear identification of the license plate. In unattended monitoring, equipment needs long-term stable operation under unattended conditions.

Outdoor installation, unattended. In the field of security, most of the equipment needs to be installed outdoors, and the equipment needs to withstand the wind and sun for many years. It is subject to the influence of natural conditions such as temperature, humidity, salinity and radiation. The aging of electronic devices themselves will be relatively faster than in other fields. Aging of cameras, electronic devices, transmission lines, etc. can cause images to become increasingly blurred.

The demand for a large number of video channels. In the large-scale safe city monitoring project, the number of video channels will reach tens of thousands or even more. Such a large number of video channels puts high demands on network bandwidth and storage devices. Therefore, in the field of video surveillance, it is expected that the code rate compression ratio of video coding reaches a high level, thereby reducing the bandwidth and capacity requirements. This leads to more information loss in the video coding process, resulting in

Blurred image

These special application places of security will lead to the decline of image clarity, which in turn has high requirements for image clarity, which will inevitably lead to a broad application prospect in fuzzy image processing technology.

Video surveillance image processing technology development trend

I also introduced intelligent video analysis and wide dynamics in the technical application. There are still some shortcomings or development space, such as improving the accuracy and focusing on solving the processing speed. The huge amount of information and data volume and processing speed are still a pair. The main contradiction. In addition, video surveillance image processing technology due to lack of performance of some devices, objective conditions and other factors, in the actual video surveillance applications, there will still be problems such as blurred video images and incapable key information capture. In the process of video image processing, due to operational technical problems or objective factors, it brings some negative effects to the application of video image processing technology, and reduces the level and quality of processing technology. Therefore, we are very meaningful in exploring the future development of this technology. There are several trends as follows:

First, in the era of big data, future data image technology emphasizes the development of high-definition, high-speed transmission, real-time image processing, three-dimensional imaging or multi-dimensional imaging, intelligence, and automation.

Second, in terms of intelligence, we strive to make computer recognition and understanding work in accordance with human cognition and thinking, and can take into account subjective and non-logical thinking. One point to note here is that AI can do image data and content extraction and analysis in terms of intelligence, but it can't make a fuss about video image processing.

Third, the future video image processing technology emphasizes the convenience of operation and application, and the centralized trend of image processing functions is inevitable. Therefore, the development of hardware chips is very important. For example, Thomson has designed the Systolic structure to solidify many functions of image processing on the chip.

Conclusion

Image processing technology originated in the 1920s, when the first digital photo was transmitted using digital compression technology from London, England, through New York. Over the years, video image processing technology has been widely used by the security industry, and technology updates are changing with each passing day. If image processing is treated as a detail, then the details will determine success or failure, as will the monitoring industry. (Author: Luo Chao)