CN114299107A - A multi-photoelectric intelligent tracking method in a water area detection system - Google Patents

Abstract

The invention discloses a multi-photoelectric intelligent tracking method in a water area detection system. The existing intelligent tracking scheme is divided into two parts: one is to identify the ship through the intelligent ship identification device of the central station and extract the feature vector A and send it to the photoelectric tracking controller , through the intelligent identification computing power distribution algorithm, the intelligent identification computing power can support as many optoelectronic devices as possible; the two optoelectronic smooth tracking controller receives the video signal, obtains the moving target feature vector through the ship intelligent identification algorithm and matches it with the A vector, triggering The KCF Tracker tracking algorithm tracks the target. The two parts work together to reduce the lag and delay caused by the photoelectric device when it continuously tracks the target for a long time, and ensure that the photoelectric device can accurately track the ship.

Description

A multi-photoelectric intelligent tracking method in a water detection system

technical field

The invention relates to the technical field of radar detection, in particular to a multi-photoelectric intelligent tracking method in a water area detection system.

Background technique

The most important function in the water area radar-photoelectric detection system is to use the radar to detect the ship and guide the photoelectric equipment to track the ship for a long time. At present, the long-term smooth tracking of target ships through artificial intelligence algorithms is widely used in water area radar-photoelectric detection systems. Common intelligent tracking solutions on the market are: 1. Radar guides photoelectric aiming at the target; 2. Photoelectricity sends the image to artificial intelligence equipment, intelligently identifies the ship, and calculates the pixel displacement vector from the center of gravity of the ship to the center of the image; 3. According to the radar coordinates of the ship , The coordinates, rotation angle, pitch angle and other information of the current photoelectric device convert the pixel displacement vector into the photoelectric PTZ control signal. Automatically control the photoelectric device to smoothly track the target, keep the target in the center of the screen and zoom in to an appropriate size.

However, the artificial intelligence recognition software and hardware in the target water area detection system directly use the mature algorithms and hardware equipment in vehicle recognition and pedestrian recognition, and have not thoroughly studied the unique characteristics of ship tracking and recognition in water area detection, resulting in existing intelligent recognition. Software and hardware devices do not work well for water detection.

In the existing water detection system, it is often a radar with 4-5 heavy-duty pan-tilt photoelectric devices. Both radar and optoelectronics are deployed in the field, where power supply and network bandwidth are extremely limited. The central station is deployed in the urban computer room away from radar, optoelectronics and other detection equipment. If the artificial intelligence identification software and hardware are deployed at the central station, the network delay will cause a large delay in the control signal sent by the intelligent identification software, and the photoelectric equipment cannot accurately track the ship. If artificial intelligence software and hardware are placed on the optoelectronic equipment side, the artificial intelligence software and hardware consumes a lot of energy, and the power that can be provided in the field is usually very tight and cannot support high-power equipment. And artificial intelligence software and hardware are expensive to add, and the cost of each photoelectric device equipped with artificial intelligence equipment will greatly increase.

In addition, ships usually sail very slowly, and the monitoring radius of heavy-duty gimbal photoelectric equipment is 10-20 kilometers. Usually, the photoelectric equipment will continue to track and monitor the same ship for more than 10 minutes, and the artificial intelligence recognition algorithm will continue to identify the same target for a long time. , which is a waste of computing power.

SUMMARY OF THE INVENTION

In order to solve the above problems, the present invention provides an intelligent identification method for multiple optoelectronic devices in a water area detection system, which divides the existing intelligent tracking scheme into two parts: one is to identify the ship through the intelligent ship identification device of the central station and extract the feature vector A and send it to the The photoelectric tracking controller, through the intelligent identification computing power distribution algorithm, allows the intelligent identification computing power to support as many optoelectronic devices as possible; the second photoelectric smooth tracking controller receives the video signal, obtains the moving target feature vector through the ship intelligent identification algorithm and combines it with the A vector Match and trigger the KCF Tracker tracking algorithm to track the target. Through the coordinated work of the two parts, the lag and delay caused by the photoelectric device when it continues to track the target for a long time is reduced, and the photoelectric device can accurately track the ship.

To achieve the above object, the technical scheme of the present invention is as follows:

A multi-photoelectric intelligent tracking method in a water area detection system, the specific steps are as follows:

S1. The radar guides the optoelectronic equipment to aim at the target vessel: if entering the warning area, trigger the radar to continuously send the target coordinates, and if the target deviates from the route, trigger the radar to continuously send the target coordinates;

S2. The intelligent vessel identification device at the central station receives the video stream sent by the optoelectronic device, and the intelligent identification algorithm identifies the target of the video stream, extracts feature points from the identified vessel and forms a feature vector A, and sends the target feature vector A;

Through the intelligent identification computing power distribution algorithm, the intelligent identification computing power can support multiple optoelectronic devices at the same time;

S3. The photoelectric tracking controller receives the target feature vector A, and tracks the target ship through the KCF Tracker tracking algorithm, and calculates the pixel distance between the target center and the video center;

S4. According to the radar coordinates of the target ship and the current PTZ information of the optoelectronic device, the optoelectronic tracking controller calculates the control signal of the optoelectronic device through the optoelectronic pan-tilt control algorithm and sends it to the optoelectronic device;

S5. The optoelectronic device adjusts the rotation angle, pitch angle, and focal length by means of the control signal, automatically controls the optoelectronic device to smoothly track the target, and keeps the target in the center of the screen and zooms in to an appropriate size.

Further, the optoelectronic tracking controller described in step S3 receives the video signal of the optoelectronic device, and obtains the moving target through the sparse optical flow method; calculates the feature vector of the moving target, and performs feature matching with the feature vector A, and uses The KCF Tracker tracking algorithm tracks the target.

Further, in step S4, the pixel displacement vector is converted into an optoelectronic PTZ control signal according to the radar coordinates of the ship, the coordinates of the current optoelectronic equipment, the rotation angle, the pitch angle and other information.

Further, the central station intelligent ship identification device sequentially identifies the video stream sent by each optoelectronic device; identifies the ship in the video, and then combines the PTZ information of the optoelectronic device and the radar coordinates of the target to mark the target ship in the video, and The feature vector A of the extracted ship is sent to the corresponding photoelectric tracking controller to help the photoelectric tracking controller "manually select the target".

Further, the photoelectric tracking controller can successfully trigger the KCF Tracker algorithm to achieve target tracking by relying on the feature vector obtained by the ship's intelligent identification algorithm; when the photoelectric tracking controller fails to track, it sends a signal to the central station to trigger the ship's intelligent identification. The algorithm "manually selects the target" again.

Compared with the prior art, the present invention has the following advantages:

1. The present invention installs an intelligent identification computing power distribution algorithm on the intelligent ship identification equipment of the central station, which can realize intelligent identification of video streams sent by multiple optoelectronic devices at a lower cost.

2. The target tracking algorithm and the photoelectric pan-tilt control algorithm are installed in the photoelectric tracking controller, which can greatly ensure the factuality and smoothness of the photoelectric pan-tilt control, and reduce the lag and delay caused by the photoelectric tracking the target for a long time. . The smooth tracking control of multiple PTZ cameras can be realized at low cost.

3. When the central station directly controls the camera, there is usually a delay of 2-3 seconds. In the present invention, there is no delay in the camera control under the condition of automatic identification and tracking.

Description of drawings

Fig. 1 is the composition diagram of the multi-photoelectric intelligent tracking method of the present invention;

Fig. 2 is the flow chart of the algorithm of intelligent identification computing power allocation;

Fig. 3 is the flow chart of the target tracking algorithm;

Fig. 4 is a flow chart of an algorithm for intelligent identification of ships;

Figure 5 is a flow chart of the control algorithm of the photoelectric pan-tilt head.

Detailed ways

Specific embodiments of the present invention will be described in detail below. It should be understood that the specific embodiments described herein are only used to illustrate and explain the present invention, but not to limit the present invention.

As shown in Figure 1-5, a multi-photoelectric intelligent tracking method in a water area detection system, the specific steps are as follows:

S1. The radar guides the optoelectronic equipment to aim at the target vessel: if entering the warning area, trigger the radar to continuously send the target coordinates, and if the target deviates from the route, trigger the radar to continuously send the target coordinates;

S2. The intelligent vessel identification device at the central station receives the video stream sent by the optoelectronic device, and the intelligent identification algorithm identifies the target of the video stream, extracts feature points from the identified vessel and forms a feature vector A, and sends the target feature vector A;

Through the intelligent identification computing power distribution algorithm, the intelligent identification computing power can support multiple optoelectronic devices at the same time;

S3. The photoelectric tracking controller receives the target feature vector A, and tracks the target ship through the KCF Tracker tracking algorithm, and calculates the pixel distance between the target center and the video center;

S4. According to the radar coordinates of the target ship and the current PTZ information of the optoelectronic device, the optoelectronic tracking controller calculates the control signal of the optoelectronic device through the optoelectronic pan-tilt control algorithm and sends it to the optoelectronic device;

S5. The optoelectronic device adjusts the rotation angle, pitch angle, and focal length by means of the control signal, automatically controls the optoelectronic device to smoothly track the target, and keeps the target in the center of the screen and zooms in to an appropriate size.

Specific examples:

In the present invention, the existing intelligent tracking scheme is divided into two parts: the central station intelligent vessel identification device and the photoelectric smooth tracking controller.

As shown in Figure 2, the intelligent ship identification device at the central station is responsible for identifying ships, extracting feature points from the identified ships and forming feature vectors, and transmitting the ship's feature vector A to the photoelectric smooth tracking controller through the Internet. Through the intelligent identification computing power distribution algorithm, the intelligent identification computing power can support as many optoelectronic devices as possible.

As shown in Figure 3, the photoelectric tracking controller receives the photoelectric video signal, and obtains the moving target through the sparse optical flow method. Calculate the feature vector of the moving target and perform feature matching with the A vector. For the successfully matched targets, the KCF Tracker tracking algorithm is used to track the targets.

According to the radar coordinates of the ship, the coordinates of the current optoelectronic equipment, the rotation angle, the pitch angle and other information, the pixel displacement vector is converted into an optoelectronic PTZ control signal. As shown in Figure 5, the photoelectric device is automatically controlled to track the target smoothly through the photoelectric pan-tilt control algorithm, and the target is kept in the center of the screen and enlarged to an appropriate size.

The invention installs the target tracking algorithm and the photoelectric pan-tilt control algorithm in the photoelectric smooth tracking controller. The advantage of this scheme is that it can greatly ensure the factuality and smoothness of the control of the photoelectric PTZ, and reduce the lag and delay caused by the photoelectric continuous tracking of the target for a long time. However, in order to control the cost, reduce the power, and adapt to the high-humidity field installation environment, the photoelectric tracking controller needs to choose a fully enclosed small industrial computer.

Due to the limitation of hardware, the target tracking algorithm needs to choose the KCF Tracker tracking algorithm with small computational complexity and high speed. Like other non-intelligent tracking algorithms, the KCF Tracker tracking algorithm has the following defects: 1) The target needs to be manually selected; 2) The tracking success rate will decrease rapidly with the movement of the target causing the change of the shooting angle and the target being partially or completely occluded.

In order to solve these two problems, the central station of the urban command center will identify the video stream sent by each optoelectronic device in turn through the intelligent identification algorithm. Identify the ship in the video, and then combine the photoelectric PTZ information and the radar coordinates of the target to mark the target ship in the video, and send the feature vector A of the extracted ship to the corresponding photoelectric tracking controller to help the photoelectric tracking controller "manually". Choose a target".

The communication between the central station and the photoelectric tracking controller usually has a delay of 2 or 3 seconds (the distance is too far, the network bandwidth is limited), and the characteristics of the ship will not change for 3 seconds because the speed of the ship is usually very slow. As shown in Figure 4, the photoelectric tracking controller can successfully trigger the KCF Tracker algorithm to achieve target tracking by relying on the feature vector obtained by the ship intelligent identification algorithm. When the photoelectric tracking controller fails to track, it sends a signal to the central station to trigger the ship's intelligent identification algorithm to "manually select the target" again.

The foregoing has shown and described the basic principles, main features and advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited by the above-mentioned embodiments. The above-mentioned embodiments and descriptions describe only the principles of the present invention. Without departing from the spirit and scope of the present invention, there are various Variations and improvements are intended to fall within the scope of the claimed invention. The scope of protection claimed by the present invention is defined by the appended claims and their equivalents.

Claims (5)

1. a multi-photoelectric intelligent tracking method in a water area detection system, is characterized in that, concrete steps are as follows: S1, radar guidance photoelectric equipment is aimed at target vessel; S2. The intelligent vessel identification device at the central station receives the video stream sent by the optoelectronic device, and the intelligent identification algorithm identifies the target of the video stream, extracts feature points from the identified vessel and forms a feature vector A, and sends the target feature vector A; Through the intelligent identification computing power distribution algorithm, the intelligent identification computing power can support multiple optoelectronic devices at the same time; S3. The photoelectric tracking controller receives the target feature vector A, and tracks the target ship through the KCF Tracker tracking algorithm, and calculates the pixel distance between the target center and the video center; S4. According to the radar coordinates of the target ship and the current PTZ information of the optoelectronic device, the optoelectronic tracking controller calculates the control signal of the optoelectronic device through the optoelectronic pan-tilt control algorithm and sends it to the optoelectronic device; S5. The optoelectronic device adjusts the rotation angle, pitch angle, and focal length by means of the control signal, automatically controls the optoelectronic device to smoothly track the target, and keeps the target in the center of the screen and zooms in to an appropriate size. 2. The multi-photoelectric intelligent tracking method in a water area detection system according to claim 1, characterized in that, in the photoelectric tracking controller described in step S3, a video signal of the photoelectric device is received, and a moving target is obtained by a sparse optical flow method; Calculate the feature vector of the moving target, and perform feature matching with the feature vector A, and use the KCF Tracker tracking algorithm to track the target for the successfully matched target. 3. The multi-photoelectric intelligent tracking method in a water area detection system according to claim 1, is characterized in that, in step S4, the pixel is displaced according to information such as the radar coordinates of the ship, the coordinates of the current photoelectric equipment, the rotation angle, the pitch angle, etc. The vector is converted into the photoelectric PTZ control signal. 4. The multi-photoelectric intelligent tracking method in a water area detection system according to claim 1, is characterized in that, the said central station intelligent vessel identification device sequentially identifies the video stream sent by each photoelectric device; identifies the vessel in the video, Combined with the PTZ information of the optoelectronic equipment and the radar coordinates of the target, the target ship is marked in the video, and the feature vector A of the extracted ship is sent to the corresponding optoelectronic tracking controller to help the optoelectronic tracking controller "manually select the target". 5. The multi-photoelectric intelligent tracking method in a water area detection system according to claim 4, wherein the photoelectric tracking controller can successfully trigger the KCF Tracker algorithm to achieve target tracking by relying on the feature vector obtained by the ship intelligent identification algorithm. ; When the photoelectric tracking controller fails to track, it sends the signal to the central station to trigger the ship's intelligent identification algorithm to "manually select the target" again.

Images