Panoramic camera surveillance location and classification technology

Panoramic camera application and classification technology

This article refers to the address: http://

With the improvement of people's security awareness and the rapid development of security technology, video surveillance is becoming more and more important according to the role played in security systems. Users not only need to see the images clearly, but also hope to see more extensive monitoring. The scene, therefore, the need for panoramic surveillance is presented.

Monitoring site requiring a panoramic camera

According to the actual experience of participating in the project, the main applications of panoramic monitoring are usually the following.

1. It is only necessary to monitor the entire situation of the entire site, and do not need details, such as a wide range of public places, traffic pivots, traffic intersections. Due to the wide view of these places, the width is even more than 100 meters. While seeing the panoramic image to achieve scheduling, it is impossible and no effort to see the details of the vehicle license plate number. In these occasions, it is only necessary to install a camera at the commanding height to realize panoramic monitoring. At this time, a fisheye camera can be used to meet the panoramic monitoring in one place.

2. At the same time, the monitoring site with special attention to detail can realize the monitoring without dead angles, such as banks, supermarkets and shopping malls. In places where robbery and theft of security incidents are prone to occur, the suspects’ faces should be seen as fine as possible. This requires not only the panoramic process of monitoring the entire crime, but also the characteristics of each person that are different from others.

3, not only need to monitor the overall situation, but also highlight the special circumstances, early warning to capture the scene of the event, so that the basic functions of the panorama while achieving the focus of the effect, but also a feature, while achieving monitoring effects and costs Take care of both. The low-resolution panoramic camera monitoring is adopted, and the built-in monitoring behavior algorithm software is combined with the automatic PTZ camera to realize the basic functions of panoramic monitoring and key image capture. Through the software calculation and analysis of the panoramic video screen, if a preset video screen appears, such as a sudden emergence of moving targets in the no-man's land, people running on the square, vehicle accidents, and close-up cameras with automatic pan/tilt are closely monitored by software operation.

4. There is also a so-called panoramic monitoring that is best understood. In order to monitor the entire scene, installing multiple cameras from different angles is a budget for the user and therefore not economical. At the same time, the video images captured by multiple cameras have a certain difficulty for the continuous monitoring of a certain moving object within the monitoring range due to the poor correlation between the pictures.

Panoramic camera classification technology

The panoramic camera can be divided into two types, one is a special super wide-angle fisheye lens with a single sensor, and the fisheye panoramic camera that relies on image correction technology to restore the image. Excellent cost performance makes these products occupy the mainstream market share, but the particularity of the fisheye lens will cause a waste of sensor pixels, and the distortion of the edge of the picture is difficult to achieve high definition, which limits the definition within the monitoring range. Coverage area; another product is composed of multiple sensors combined with special lens to realize the panoramic function. This kind of multi-lens stitching panoramic camera does not need corrective operation because each sensor gets a regular rectangular image, but the corresponding needs another The algorithm software can realize the seamless splicing of the picture, and the demand for the whole set is high, that is, the setting of the lens angle of view and the installation position are strictly required. And because the hardware uses multi-lens and multi-sensors, such products are difficult to control costs, and the price is generally higher than the fisheye panoramic camera.

Fisheye panoramic camera

Primary stage: The product only has the function of outputting panoramic images and quad-screens;

Intermediate stage: The output image can perform some deformation correction and virtual PTZ, but the deformation and correction effect is not good, and the lower resolution (less than 3 megapixels) makes the image processing after the resolution is not good, and The use of traditional compression coding algorithms also makes video processing too consuming the resources of the PC server;

Advanced stage: The product resolution can reach more than 3 megapixels, and it can flexibly perform image deformation correction and virtual PTZ as needed.

The structure of the fisheye panoramic camera is not complicated, only the fisheye lens and image correction software. The fisheye lens is famous for its front end as a fisheye. The focal length of the fisheye lens used in the panoramic camera must be less than 2mm. Currently, such lens products are relatively mature on the market. The reason why the fisheye panoramic camera becomes a high-end technology product lies in the difficulty of the original distortion pie chart processing and how to better provide the humanized service for the back-end monitoring personnel, that is, the realization of the two core technologies of image correction and virtual ptz. It can also be seen from the above that it is these two technologies that determine the development level of the fisheye panoramic camera.

Specifically, as long as there is a fisheye lens, a panoramic picture can be obtained in the background, but the original image is a distorted circular pie chart that does not conform to the normal observation habits of the human eye and must be corrected to achieve the final two-screen or four. Split screen monitoring needs. The virtual PTZ technology gives the user a monitoring experience like a fastball, which realizes the key observation, image rotation and amplification of any monitoring point within the observation range without mechanical components.

The idea of ​​image correction technology is not complicated. Generally, a single fan shape is drawn from the original pie chart, and then certain deformation and correction processing are performed according to the principles of proportional and perspective, and the separately processed fan images can be combined to meet the monitoring requirements. image. For products that are not in high demand, the circular pie chart can be directly stretched in a simple four-corner shape, and an image effect suitable for human eyes can be obtained. Although the core algorithm of each manufacturer is bound to be different, the central idea of ​​image correction technology is to use a certain algorithm to integrate the graphic distortion with severe edge distortion into a normal scale image suitable for viewing. The rationality of the algorithm used, the level of programming and the final system resource occupancy rate jointly determine the pros and cons of this algorithm. As for virtual PTZ, it is actually performing pixel capture inside the image to achieve functions such as zooming and rotating. The main difficulty of the contrast correction algorithm is not how to achieve better image effects, but how to better integrate it into the front-end firmware. Two-way virtual control of facts and post-recording is implemented with the back-end software.

Diverse panoramic surveillance

According to the specific location and mode of implementation of these core technologies, the current fisheye panoramic camera can be further classified into the following three categories:

1, software front-end firmware integration (software mode)

All of the camera's functional implementation software, including image correction, virtual PTZ, smart components, and compression algorithms, are all integrated into the firmware inside the camera, and future upgrades require only the front-end settings to be updated. At the same time, its background PC can also provide virtual PTZ and other functions, giving users the possibility to flexibly change the viewing angle and focus afterwards. And because the same amount of computation differs greatly in the amount of load on the hardware crystal and the software PC (the same computing task, when it comes to the problem of computing matrix, it will occupy a lot of software resources, the operation on the hardware crystal is relatively faster.) Therefore, this front-end integration greatly releases the resources of the back-end server and the network. Combined with distributed architecture, such products can minimize the configuration of back-end servers when applied to large and medium-sized projects. Generally, even large projects need only be configured with a PC-side server, such as DAS, NAS, and so on. On the other hand, due to the technical limitations of the front-end firmware, it is expected that such panoramic cameras can only process up to five megapixels. With the advancement of high-definition, whether this bottleneck can be broken will be the biggest development problem for such panoramic cameras in the future.

2, chip front-end implementation (hardware mode)

Nowadays, Taiwanese manufacturers have developed chip products that can directly realize basic functions such as image correction and quadrant output. They are also interested in fisheye panoramic cameras, which can be directly integrated with such chips and fisheye lenses. product.

All the functions in this mode are also implemented in the front end, but unlike the software front-end firmware integration, it does not allow the retention of the original image and subsequent operations. After the user selects the inherent function, the machine processes the picture inside the front camera according to the instruction, and then outputs to the back end to realize the split screen or quad screen effect.

The emergence of this method has given some manufacturers who do not have independent research and development capabilities to "close contact" with panoramic cameras, and also applies to a certain range of low-end markets. However, all its functions are packaged in the finished chip, it is difficult for the manufacturer to further upgrade itself, or to develop new application functions, and it is not allowed for the user to change the viewing angle according to actual needs after the event, or perform corresponding analysis and processing operations, which has great limitations. Sex. The urgent task for the future is to increase the flexibility of this type of product and consider whether space for rewriting and upgrading can be reserved inside the chip.

3, back-end software implementation

All functions of this type of panoramic camera are implemented by software installed on the back-end PC. The front-end camera is only responsible for capturing images, so the product has advantages in cost and price.

Relying on the powerful processing power of the back-end PC, such products do not have the bottleneck limitation of the definition upgrade. While implementing the functions of basic multi-segment display, virtual PTZ, etc., the preservation of the original image also enables more information collection and later. Processing work. On the other hand, such products have high consumption of back-end PCs and network resources, and need to be configured with more back-end servers for large projects.

Multi-lens panoramic camera

Such a panoramic camera is internally packaged with a plurality of sensors, and a panoramic effect is obtained by performing image stitching operations on the divided screens. The current mainstream product is structured by encapsulating four two-megapixel sensors and a separate short-focus lens with an angle of view of 45° or 90° in a uniform housing. The core technologies such as digital processing and compression are integrated on the front-end firmware, and the four separate pictures are integrated into a 180° or 360° high-definition panoramic picture according to user requirements, and then transmitted to the back-end management platform by the network. Compared with the mainstream fisheye panoramic camera, the advantage of the multi-lens stitching panoramic camera is that it is freed from the limitation of the focal length to some extent, and the farther distance can be seen under the same conditions. At the same time, it also has a virtual PTZ function, which can perform functions such as peripheral cropping, intermediate restoration, and 180° expansion, which is convenient for users to adjust the screen according to the monitoring focus.

Compared with the current mainstream fisheye panoramic camera, the multi-lens stitching panoramic camera has no pixel limitation, and there is no such problem that the difference between the middle and the edge is huge. Specially selected chips for the security industry, completely abandon redundant algorithms and space waste that are not related to security, and the underlying algorithms are self-developed. Therefore, the image stitching of the camera, although the compression is implemented at the front end, but its speed and network resources are very small, which will be the future development trend of this type of product.

Due to the use of multiple lenses and sensors, such products are more expensive than fisheye panoramic cameras, but still cost less than directly installing multiple conventional cameras, and they only have one IP address, which is convenient for software licenses to solve problems. On the other hand, although such products avoid the waste on the pixels, the splicing technology has higher requirements for the whole set of solutions, in which the choice of the viewing angle, the structure of each other perfectly matched, how to better seamlessly splicing the algorithm, All decide whether such products can achieve better development in the future. Moreover, because the cooperation of the lens can not achieve physical perfection, it is impossible to avoid a conical blind spot directly under the lifting. After improvement, the camera's internal lens pitch angle is adjustable, and the impact of this blind zone can be reduced to some extent according to the site conditions. On the other hand, when the installation height reaches more than three meters (the optimal installation height is three meters to five) m), the size of the blind zone generated by the camera is basically similar to the size of the human head, that is, it does not have much influence on the actual monitoring.

In addition to the basic lens and core technology, the quality of a panoramic camera depends on its choice of sensors. If the pixel of the selected sensor is too low, it will directly affect the final display effect of the corrected and stitched picture. On the contrary, if the sensor with too high pixel is selected, it will cause excessive computational pressure on the hardware. And according to different design ideas and product positioning, some manufacturers that mainly monitor indoors do not design infrared monitoring functions, while others choose dual lens or electronic conversion to achieve nighttime monitoring.

Because of the huge amount of information acquired by itself, coupled with software correction, splicing technology and virtual PTZ occupancy of resources, the panoramic camera has more resources for the network and the back-end system than ordinary cameras. The front-end processing method has little pressure on the network and the back-end, but the corresponding front-end resource consumption is huge, and the product price is also high. On the contrary, the implementation of the back-end software tests the tolerance of the entire network and the back-end server. Although the individual products are cheaper, the supporting service devices need to be increased accordingly. Therefore, in the early planning of the project, the entire network and project capacity should be taken into consideration, rather than considering the investment of the camera alone. The distributed architecture is currently recognized as the most suitable for panoramic camera projects.

5.5inch PDA Handheld Terminal

Pda Rfid Scanner,Pda With Fingerprint Scanner,Pda With Built In Scanner,Android Handheld Terminal With Printer

Shenzhen Qunsuo Technology Co., Ltd ,