Intelligent home security monitoring or to a turning point

Home security is inseparable from video surveillance systems, video surveillance system has also been widely used in residential security monitoring, fire security monitoring, traffic violations, traffic control, and military, as well as banks, shopping malls, subways, airports and other public places, and It will have a wider range of applications.

However, the existing video surveillance systems typically can only be done by manual monitoring, alarm, recording a series of analysis processing, not fully real-time active monitoring role. More product demand and the broader market applications require future security products must have the ability to intelligent analysis process. From the current level of technology, the digital image processing technology has become more sophisticated, artificial intelligence by leaps and bounds, how intelligent image analysis functions into smart products to become each company and focus of manufacturers. With advanced intelligent analysis of intelligent video surveillance system, not only can greatly enhance the ability to monitor and reduce the hidden dangers, while saving manpower and resources, reducing costs.

Intellectual development in the monitoring field

In the early 1990s prior to simulate equipment as the main CCTV system is the first of the first-generation monitoring system, the technology is mature, stable, and in many practical projects have been widely applied in some cases This technology vision which are still of concern. The real intelligent monitoring and innovation with the development of computer networks only appeared. The mid-nineties, the development of computer processing power increase and video technology allows people to start using the computer's high-speed data processing capabilities for video capture and processing arises digital surveillance system called the second generation of video surveillance systems . From the late 1990s, with the advent of network bandwidth and computer processing to further improve the power and storage capacity, as well as various video processing technology, video surveillance into the new digital age. Intelligent and more important in the security of the application. So, is a web-based, digital video compression, transmission, storage and playback as the core, intelligent image analysis technology revolution featuring third-generation surveillance system. Intrusion alarm, illegal residence, items left behind, moving target tracking, and so on. These advanced intelligent analysis capabilities to become a modern security monitoring products, a major feature.

Research related to intelligent algorithms

Mention must mention the intelligent security, mention must mention intelligent algorithm. Soul algorithm, as behavior analysis and image processing, and acts as a guide in the future of product development and program implementation. Field monitoring of continuous expansion, continued development of the market, making huge video information needs analysis process, for intelligent video surveillance system, the face of such a large amount of information, technology and good structure is the key problem. In the remote monitoring process, in order to reduce the pressure on the network transmission, in front of the device some intelligent analysis module embedded coding module to which product to go through the front of the intelligent video analysis is only extract useful information, and encode compression and transmission , greatly reduces the amount of data transferred, and the image analysis part in the decoding process by the rear end portion of the software. This division of labor type thought to be better than the effect simply by centralized processing software or hardware.

Compared with H264 codec algorithm, etc., intelligent video algorithms is no reason why a uniform standard, no ASIC technology, because of its complexity and the diversity of the target behavior, makes it difficult to develop a standard set of rules to be applied to each industries. Intelligent algorithm is an emerging class of integrated science and technology, many effective intelligence analysis algorithms are developed first from abroad, but also a significant portion of the complexity of the algorithm is either too high, not suitable for real-time requirements; or adaptive poor , too much dependence on the scene; or are still in the phase of theoretical research is still relatively far away from the product. Domestic completely independent intellectual property rights truly intelligent monitoring products rarely, most technology products are from the United States, Europe, and Israel and other countries. Most domestic production and installation of video surveillance systems still remain in the field of non-intelligent, mentioned in the monitoring system "smart video surveillance", in fact, still in the concept of common network video surveillance.

Most or IP surveillance and digital surveillance, etc., do not have the advanced intelligent analysis of said content. For this reason there are many companies in the domestic intelligence also put a lot of manpower and resources to research and development, and in the tradition of moving object detection has done a lot of improvements, and achieved breakthrough results. With independent intellectual product of intelligent security products, to improve the country's overall level of intelligent control, improve the status of domestic products in the international competition has a very important role.

From the algorithm point of view, there are a lot of intelligent monitoring products in the technical aspects of the research to be done, but it also gave us an opportunity, who mastered the core technology in the future development, we will be able to dig first pot of gold, accounting for market-related fields. Intelligent image analysis, moving target detection and tracking should be the most basic function of advanced intelligent analysis inside. Is the basis for intrusion alarm, illegal stay, dropping relics behavior analysis. Following a brief analysis of the object recognition and tracking process.

Specific application examples

Implementation of the algorithm can be seen as a process of modeling can also be understood as a kind of applied mathematics project. Moving target detection and identification means separate from the background from the foreground image in the sequence change is a major part of the digital image processing. Changes in weather and light, complex background interference of other objects, shadow and movement of the camera movement, shaking and so will target extraction process difficult. So moving target detection and accurate tracking of the right to become a video surveillance system, an important technical point. The whole process can be simply divided into the following steps: video pre-processing, object detection and recognition, object classification, object tracking, behavior analysis and rule creation.

Video preprocessing

Video pretreatment process is actually to reduce noise, outstanding track the target. There are two types of image noise, one is a source of noise, one belonging to observe noise. Changes in ambient light, background, leaves, snow and so on are a source of noise; noise is observed mainly refers to the camera shake, affect the transmission line interference. The noise will be on the tracked object extraction process to cause trouble. In terms of the general methods used to eliminate noise digital image stabilization, adaptive learning background, thresholding, morphological filtering, binarization. These processes can eliminate most of the noise in the image, reduce its interference to target detection.

Target Detection and Recognition

After using the motion object extraction process has been moving foreground objects, we need to have a target identification process. This identification into a single goal of doing more to identify and target areas identified. For tracking individual objects, it can be selected from an object of interest in the video man, and track; for the region for more than an object, you can divide a region, only the object enters the area of ​​identification and tracking.

Target classification

For a single object tracking and monitoring in the region it is not required classification. Even if the object occlusion occurred in the course of the campaign, when the object under linear movement, change the background relatively flat, the algorithm can still be accurate tracking of objects. For non-linear motion of objects in complex environments, and we are in consultation - to add some single target feature information on the basis of the target tracking to help identify track. Target classification generally used in the case of multi-target tracking under a single target, or multi-target tracking multi-target environment. When there are multiple moving objects in the environment when we can put the objectives of each campaign are extracted, according to position, speed, shape, texture and color and other relevant characteristics to classify them. First, save the characteristic information of interest to the target down, then be tracked. When the video track objects disappear and then appear, we will re-match with all the moving targets before the advent save feature information and the current picture. By comparing the results, it can determine which one object appears is just tracking the goal.

Behavior analysis and rule creation

If the object tracking process is a purely technical process, then it should belong to advanced behavioral analysis of the scope of the intelligence. Simply put, that is the basis of human behavior analysis rule creation, based on the video images of the automatic analysis and extraction of video source in a process critical information. Because creating rule is related to the environment, such as the scene is different, set the trigger alarm conditions and different regions. Therefore, the development, the software side of the focus of the work in intelligent security analysis is to try to make the smart move humane direction, in simple settings, you can make intelligent analysis of the entire image has a self-learning process of adaptation, further reducing human the impact of factors, alleviate human workload.

Summary and Outlook

With the increasingly rapid development of China's security industry, the market is expanding, a company engaged in the security industry is also increasing. For the promotion of intelligent security products company, only to seize the pulse of the security market, mastered the key technology to respond to future market needs correctly and continuous innovation, constantly find opportunities in the market competition.

Compared with foreign countries, China's intelligent security technology is still a long way to go, but it is certain that with the frequent increase in the number of domestic and international technology exchange, doubling the number of domestic talent and improve the development environment, these gaps will be more be smaller. Hope that the various domestic companies in the security industry can protect the company in the context of intellectual property rights, a lot of communication and exchanges, and constantly improve the security industry specifications and market norms. Under the premise of establishing a win-win market, continuous technological innovation, with intelligent image processing technology as the core of intelligent video surveillance products widely in various industries and fields, creating a peaceful and safe environment for people's lives.