Image motion sensors usually and monitoring devices used in combination, once the camera input images have unusual sound will be immediately issued a warning. In the ideal state, only on the monitor screen found suspected of theft, robbery, and so on and so forth, the alarm system will notify the relevant units.

Traditional image motion detecting system adopts analog signals to detect, through surveillance camera input image brightness difference distinguish between each image block, if brightness is different from the original set, identify the time the alarm will start.

Such products are still being in circulation, but mainly indoor application. Outdoor detection distance, and light environment is not fixed, and the birds, for example, cloud caused by camera shake, trees, the wind is too high, even the heavy rain, snow and ice, such as climate, can cause false positives, so this kind of product is not suitable for outdoors.

Therefore, in order to cope with difficult outdoor environment, we need more sophisticated image processing technology, and, thanks to the development of digital technology vendors can offer more accurate imaging detection products.

Although digital image motion detecting system is also to "identify a specific period of time, the brightness of the image set scope change" as the principle, but digital system emphasized for determining the intruder action (direction, location), and the differences of landscape lighting. What is more, the latest equipment have learning function, can automatically determine the background change of state, joined the environment variables in order to enhance accuracy.

Measuring the program

Outdoor image detection measurement step, cutting into discrete part is the image, determine the image changes of the light part in every period of time, and then tell whether these changes resulting from the conversion or moving objects.

In general, has the following kinds cutting images

- which segmentation (pixel division) : image segmentation intoVery small picture composition (pixels), there may be millions of blocks;

- community block segmentation (small - cell division) : the imageSeparated for hundreds or thousands of geometric size or different sizes block;

- big block segmentation, - cell division) : the imageDivided into hundreds of different size of the block.

Due to the pixels of shredding will get quite a lot of image information, thus requires a lot of resources to identify computer calculation. And although the way of big block segmentation will not cause the burden of computer resources, but due to provide the image information is less, so it is not easy to distinguish between the change of the environment. Therefore, adopts the neighborhood block segmentation method can get enough image identification information, don't worry cause computer calculation load, is the most appropriate way to distinguish.

The differences of the light change

In order to achieve "only when the object moves to produce alarm" the goal of the system must identify "moving objects" and "natural weather", "the camera aperture adjustment" the light of the differences.

During outdoor image detection, must consider the scene image illumination is enough, and the light intensity and changeable. This means that when the invaders into the photography picture, if at that time, dimly lit, moving images of light and shadow is not obvious; However if the light is enough, or has a better light environment, changes of the light can be clearly moving images. If the image motion detecting compared with preset light only differences in values, as the basis of found an intruder, are insufficient light place will not be able to use, and the bright place will be overly sensitive.

Part of the system has flexible adjustment function, make installation project, it can be set on the basis of field environment changes of the light alarm when sending the parameters. But when the sun changes position, and makes the monitor screen location of the building, the shadow of street trees change position, make the function can't effectively play a role in the daytime.

Good practice, is computing scene contrast change, and whether these data identify as invaders. For most of the scene, contrast change produced by the invaders moving with the scene of light and shadow changes are of two different parameters.

In order to provide high accuracy of outdoor image motion detecting function, must control all kinds of outdoor environment changes, including intensity of illumination, and environmental factors such as light source transformation. Due to image illumination change constantly, so as long as the different change image blocks, simple type of image motion detecting system will trigger the alarm, and produce false positives. And more complex system will vary according to the image motion of the overall difference, in order to maintain the best state of detection and reduce the false alarm.

Recognition program

On the market at present, there are two relatively commonly used identification method, an object is size detection, and then looked at mobile process over a period of time and space, to infer continuing mobile location.

Effective object size discrimination can be dramatically reduced because of the small animals or false positives, produced by wind blowing from simple type image motion detecting equipment faults. In addition, the object motion detecting pay special attention to: when the invaders is close to the TV camera, object size will bigger than stay away from the camera.

Pixels split type measurements provide the standard size object recognition. Identify objects dimension technology, can provide a measurement program according to standards. Measuring steps from the calculation of total pixels to individual brightness change affected by the group of paintings, panorama, the invaders to render the possible images.

Simple use of shadows and size information, may be because the monitor screen appear the size of a human animal and trigger the alarm. Therefore, this approach may be too simple and easy to be cracked. But from a practical application, the smallest object size measurement and visual identification collocation operator application condition and behavior observation, which can effectively distinguish cause alarm.

Mobile identification

Even strengthen the function of the various complex, unable to effectively identify the scene such as the electric light open and the light change between mobile target, the image motion detecting still unable to provide the user a good outdoor application effectiveness. Therefore must establish a set of records in accordance with the time course of operation mode, to correctly pointed out that whether for moving objects.

Simple image motion detecting system using the default time information, the computing system can improve the single events record of bad points, but does not provide mobile evidence, and are easy to be fixed by flash or other ways to cheat and produce false positives.

Other systems can allow users to set several warning and alarm at the same time zone. Area to detect sequence is warning area of objects at first, and the location of the objects in the alert zone within the given time; So, can learn the movement of the object position. However, if want to get good results, the system often need to carefully set, rather than a simple use of the system.

Need higher difficulty of calculating program is to use an object tracking algorithm. By these complex program, to understand the relativity between the object size, to estimate the moving process is regular interference or temporary emergencies.

In fact, a good set of multiple object tracking system should be able to distinguish between shaking and moving object is different, separate each object tracking, and within the scope of the smallest mobile alerts. Even, the automatic tracking system should provide tracking after the calculation results, including the moving course of time, place, object, in order to calculate the moving velocity.

Learning technology

A new generation of image motion detecting products more to join the learning technology, to strengthen the big range detection efficiency, and set installation personnel need not possess professional skills. Learning program will continuously to applied to the changing environment, to provide better detection ability, reduce the rate of false positives.

Technology is mentioned above all rely on the measurement of the differences between each time the light, so don't need to recognize objects fixed the ability to move. Both of permanent objects in the scene, scene's own mobile or gradually moved into the new object of the image, they can by object tracking process, from a wide range of image to move or inference from the complex formation, and then trigger an alarm.

Another alternative way to restore the detected moving objects and learning architecture related picture. By the way, the detector can effectively distinguish the behavior of the non-invasive objects such as camera shake, shake, or shadow, and the actual invasion object.  

We can think so, a practical system should be able to automatically identify environmental conditions change, and can identify objects by detecting the program moving, in some special cases, can effectively reduce the rate of false positives.

Learning techniques provide better weather and environment recognition performance, whether it is a heavy snow day or fast moving clouds, etc., can be adjusted according to the environmental changes, and makes the image detection process more effectively.

Dual induction

Even if the technology is so developed, but if possible, security system should be according to the actual conditions, using more than one sensor technology, to provide better detection control, put an end to false positives.

With passive infrared detector, for example, when the temperature of the induction to the human, you can start the video motion detecting. Infrared detector by radiant infrared receiver screen, overlapping in the optical filter camera can perspective, makes the image motion detecting system in the human infrared wavelength range to obtain maximum induction degree.

Combining the two sensing technology also has the economic cost of installation. Both need only a single point of equipment installation, single point installed complementary technology both in fence system (generating systems) lack of, also provides monitoring and anti-theft function.