CN1220366C - Automatically identify and track moving objects and obtain clear image samples - Google Patents

Abstract

The invention relates to a method for automatically identifying and tracking moving objects and obtaining a sharp image, which utilizes a motion vector generated by a moving object in a digital image as a control parameter to drive an image shooting device with a motor transmission mechanism, can track an object moving into a visual field or an object moving in the visual field, can place the object into a visual field center in real time, and can provide more definite detection and search information by matching with functions of automatic focusing, visual field amplification, monitoring and warning and the like.

Description

Automatically identify and track moving objects and obtain clear image samples

technical field

The invention relates to a method for automatically identifying and tracking a moving object and obtaining a clear shadow sample, in particular to a method for automatically identifying and tracking a moving object that can utilize digital image processing technology.

Background technique

In known related applications, an image capture device (such as a CCD) is often used to monitor the surrounding environment of the target object, and/or it is matched with human monitoring. Automatic sensing devices (such as infrared thermal sensors) etc. achieve the purpose of early warning. Traditional image capturing systems mostly use methods such as fixed field of view, timed cycle cruising field of view, or manual control or active light perception to control the field of view to monitor the surroundings of the target. On a rotatable platform controlled by manipulation or active light perception. Since the defect of the fixed device adopting a fixed field of view is that the field of view of monitoring is fixed, it is easy to have the problem of dead angle.

Although the fixed-period rotation equipment can increase the wide field of vision, it still has dead spots when moving. Due to the lack of flexibility in mobile monitoring, it is impossible to accurately grasp the unexpected environmental conditions, and human monitoring is still required. And because the angle of the monitoring angle is increased, the resolution of the image is deteriorated, and it is often seen that the recorded small image is almost blurred and unrecognizable. Moreover, recording for a long time also consumes a large amount of recording media.

Devices that use manual control or active light perception to control the field of vision can make up for the grasp of sudden environmental conditions and do the latest information collection and update for environmental changes, but they also pay additional costs (such as Human or complex array signal sensing systems). Moreover, on the device, the light perception control is still unable to perform the object recognition function, and it cannot avoid human loss or negligence caused by human fatigue.

Contents of the invention

The present invention designs a method for automatically identifying and tracking moving objects and obtaining clear images. It does not need to be equipped with active light wave components (such as radio waves, infrared rays, etc.), but uses the motion vectors generated by moving objects in digital images as control Parameters, driving the image capture equipment with motor drive mechanism, can track objects moving into the field of view or moving objects in the field of view, and put the object in the center of the field of view in real time, and can be used with auto focus, magnified field of view, and video recording and warning functions to provide clearer detection and search information.

An object of the present invention is to provide an algorithm detection technology, so that the detection device can simultaneously lock multiple moving targets and select them according to self-defined rules.

Another object of the present invention is to provide a method for automatically identifying and tracking moving objects that can be applied in the fields of automatic access control, monitoring, security, etc., and can also be used in interactive toy applications as the basis of electronic pet vision to reflect the surrounding conditions at any time.

The purpose of the present invention is achieved like this:

The invention discloses a method for automatically identifying and tracking a moving object, which includes the following steps:

(a) start scanning the image;

(b) a moving object has been detected;

(c) number the object;

(d) calculate its location;

(e) Determine whether the object has moved out of the monitoring range, if so, record the data, otherwise, determine whether the object image is placed in the center of the field of view, if not placed in the center of the field of view, drive and rotate the motor transmission mechanism of each image capture device to track the object, so that Put the object into the center of the field of view, and return to step (d).

In the method for automatically identifying and tracking moving objects, in the step (a), the image can be scanned by a single image capture device with a zoom-in (ZOOM) function.

In the method for automatically identifying and tracking moving objects, the step (a) can scan images through two image capture devices, one of which is a fixed wide-angle image capture device with low price and low resolution, and the other is a high-resolution image capture device. Steering mechanism and image capture equipment with local magnification function.

In the method for automatically identifying and tracking moving objects, the step (d) can extract the operation vector of the moving object through the operation of square comparison and minimum average difference, and use the motion vector as a control parameter to feed back to the motor drive Mechanism of image capture equipment to control the lens to automatically track moving objects.

In the method for automatically identifying and tracking a moving object, the distance measurement mechanism can be used to further provide the object's auto-focus, magnified field of view, monitoring and warning information as analysis, evidence collection and observation data.

Specifically, the present invention provides a method for automatically identifying and tracking a moving object, which includes the following steps:

(a) scanning image input through at least one image capture device in a surveillance area;

(b) comparing whether there is a difference between two continuously input images to detect whether there is a moving object in the surveillance area;

(c) When a difference is generated between two continuously input images, it means that the moving object has been detected, and the moving object is numbered;

(d) Calculate the location of the mobile object in the surveillance area;

(e) Determine whether the moving object has moved out of the monitoring area, if so, record the data, otherwise, determine whether the moving object image is placed in the center of the field of view of the monitoring area, if not placed in the center of the field of view, according to the aforementioned two continuously input images Generate a difference between them, extract the motion vector of the moving object, and use the motion vector to drive the motor transmission mechanism of the image capture device to track the moving object, place the moving object in the center of the field of view, and record the image; and

(f) Notify through a notification system when the mobile object is within the surveillance area and after it has been identified.

That is, the present invention provides a method for automatically identifying and tracking moving objects that can utilize digital image processing technology, which utilizes digital image processing technology to extract moving objects from a sequence of continuously captured images through a square comparison minimum average difference operation. The operation vector of the object, and use the motion vector as a control parameter to feed back to the image capture device with the motor transmission mechanism to control the lens to automatically track the moving object. At the same time, it can be used with the distance measurement mechanism to further provide the object with automatic focus Detailed information such as magnification is used as data for analysis, evidence collection and observation.

In order to further understand the features and technical contents of the present invention, please refer to the following detailed description and accompanying drawings of the present invention. However, the accompanying drawings are for reference and illustration only, and are not intended to limit the present invention.

Description of drawings

Fig. 1 is a figure of an embodiment of the present invention;

Fig. 2 is a structure diagram of the present invention;

Fig. 3 is a flow chart of the present invention;

Fig. 4 is another embodiment figure of the present invention;

Fig. 5 is an explanatory diagram of the block comparison rule of the present invention.

Detailed ways

The present invention is a method for automatically identifying and tracking moving objects and obtaining clear shadow samples. It utilizes digital image processing technology to extract the operation vector of moving objects from the continuously captured image sequences by comparing the minimum average difference between squares. , use the motion vector as a control parameter, and feed it back to the image capture device with a motor drive mechanism to control the lens to automatically track the moving object, and can be used with functions such as auto focus, magnified field of view, monitoring and warning, etc. to provide more clear Detect search information.

Please refer to FIG. 5 , which is an explanatory diagram of the aforementioned block comparison rule. If the coordinates of the moving object in the front image P are (i, j), then the coordinates of each point on the rear image N can be searched one by one to find ∑ _{m, n=1:8} (N(k+m, l+n)- P(i+m, j+n))/64 is the coordinate (k, l) when the minimum value occurs, and the moving vector D=(ki, lj) at this time. Among them, N(k+m, l+n) represents the image brightness value of the (k+m, l+n) coordinate point in the N image; P(i+m, j+n) represents the image brightness value in the P image (i+ m, j+n) coordinate point image brightness value.

In an embodiment of the invention, the block versus minimum mean difference operates as follows:

Cut the two continuously input images (set as the front P(i, j) and the back N(K, l) images) into M*N (for example, 8*8) dot matrix blocks. When an object moves, There will be a difference between the two images in the moving area. By comparing the block of the previous image with the block with similar coordinates in the rear image, a best approximate block can be found. The coordinate difference between the two blocks represents the moving object The moving vector. The selection of the best approximation block is based on the equation ∑ _{m, n=1:8} (N(k+m, l+n)-p(i+m, j+n))/64 N when the minimum value occurs The middle square shall prevail. At this time, the moving vector (two-dimensional) is D=(ki, lj). By transforming the D vector (number of pixels) and the angle of view of the lens, correct feedback parameters can be provided to drive the image capture device to rotate to move the object into the center of the image.

Referring to Fig. 1 and Fig. 2 again, its equipment assembly comprises the image capturing device 1 with lens zoom-in (ZOOM) function, the motor 2 that rotates this image capturing device, control module 3 and notification system 4, wherein:

The image capture device 1 is a video camera with auto-focus and magnification functions.

Announcement system 4 can announce etc. by siren, light, network information or wireless telecommunication.

Control module 3 can be personal computer or microcontroller etc., and it is provided with a microprocessor 31, and this microprocessor 31 is to be connected with the output of image pickup device 1 through input port 32, and output port 33 is respectively connected with notification system 4, The motor 2 and the lens 11 are connected, and the microprocessor 31 is connected with the memory 34 .

Please refer to Fig. 3, which is a flowchart of the present invention. At the beginning, the image input 31 is scanned, and it is judged whether a moving object is detected? 32.

If no moving object is detected, go back to step 31; otherwise, carry out the identification work, that is, detect the moving object number 321, and calculate the location 322 of the moving object, and then determine whether the object has moved out of the monitoring range? 33.

If the object has moved out of the monitoring range, record the data 331, and determine whether to terminate the monitoring? 34.

To terminate the monitoring, end the job, otherwise, go back to step 31.

If the object has not moved out of the monitoring range, then determine whether the image of the object is placed in the center of the field of view? 35.

If not, drive the motor transmission mechanism to track the object 351, place the object into the center of the field of view 352, and execute step 322; otherwise, carry out identification work, calculate the size of the object 353, and then determine whether the image size is appropriate? 36.

If not, control the ZOOM function according to the size of the object, then input the image of the object 362, and execute step 353; otherwise, record the image data 363, and identify the object 364, and notify the system 365 through sound and light information, etc. Then, step 34 is performed.

Please refer to FIG. 4 , which is a diagram of another embodiment of the present invention. It can also use two image capture devices to capture images, one of which is a low-cost low-resolution fixed wide-angle image capture device 5, and the other is a high-resolution image capture device 1 with a steering mechanism and a local magnification function.

In summary, through the system equipment and software planning of the present invention, it has the following characteristics:

(1) It can automatically track moving objects, and then provide detailed information or relative responses.

(2) Eliminate expenditures for human resources or components of complex detection arrays.

(3) Using algorithm detection technology, it can lock multiple moving targets at the same time and select them according to self-defined rules.

(4) The implementation of the algorithm is flexible, and it can be quickly modified and applied to various specific occasions or special requirements.

(5) The system is compatible with other existing systems, and can also be supplemented by dual human control.

The above descriptions are only the best specific embodiments of the present invention, and the features of the present invention are not limited thereto. Any equivalent change or modification obvious to those skilled in the art can be covered by the patent scope of the present invention.

Claims (5)

1. the mobile object method is followed the trail of in an automatic identification, it is characterized in that it comprises the following steps:

(a) in a surveillance zone, import by at least one image capturing device scan image;

(b) have or not the generation difference between two images of input relatively continuously, none mobile object is arranged in this surveillance zone to detect;

(c) produce difference between two images of continuous input, expression has detected this mobile object, and with this mobile object numbering;

(d) calculate the position of this mobile object in this surveillance zone;

(e) differentiate this mobile object and whether shift out this surveillance zone, if, record data, otherwise, differentiate the center, the visual field whether this mobile object image inserts this surveillance zone, if do not insert center, the visual field, produce difference between two images according to aforementioned continuous input, extract the motion vector of this mobile object, and the drive of motor mechanism that drives this image capturing equipment with this motion vector is to follow the trail of this mobile object, make this mobile object insert center, the visual field, and recording image; And

(f) when this mobile object in this surveillance zone and through after the identification, circulate a notice of by a reporting chain.

2. the mobile object method is followed the trail of in automatic identification as claimed in claim 1, it is characterized in that the image capturing equipment of described step (a) has enlarging function.

3. the mobile object method is followed the trail of in automatic identification as claimed in claim 1, it is characterized in that described step (a) can pass through two image capturing device scan images, one of them is the fixing wide-angle image pickup apparatus of low price low resolution, and another is the image capturing equipment of high-resolution tool steering mechanism and local enlarging function.

4. the mobile object method is followed the trail of in automatic identification as claimed in claim 1, it is characterized in that the extraction of this motion vector of described step (e) can be by obtaining by the operation of square contrast minimum average B configuration difference between two images of continuous input.

5. the mobile object method is followed the trail of in automatic identification as claimed in claim 1, it is characterized in that also further comprising step: the range measurement mechanism of arranging in pairs or groups further provides the focusing automatically of this mobile object, magnification field, prison record and information warning as analyzing, searching warrant and observation data.

Images