[0041] The present invention will be described in detail below with reference to the drawings and specific embodiments.
[0042] Reference figure 1 , Is the flow chart of mine moving target tracking and positioning. First, the multi-antenna radio frequency identification device is used to identify the target and obtain its current position coordinate information, and then use this information to guide the image sensor to turn on or sleep, and use the background difference method to detect the target in the video surveillance image and perform multi-view fusion to achieve more accurate positioning Finally, the two positioning results are merged by a weighted fusion algorithm to obtain the location information of the moving target. The image sensor has an antenna module and a wireless interface function, which is used to receive communication instructions from a multi-antenna radio frequency identification device. The specific implementation steps are as follows:
[0043] Step 1. Moving target detection: Turn on the multi-antenna radio frequency identification device used for mobile target detection in the underground tunnel to keep it always on; the multi-antenna radio frequency identification device detects whether there is a moving target in the underground tunnel. If the target is detected, multi-antenna The radio frequency identification device recognizes the moving target according to its ID number, and uses the RSSI-based triangulation algorithm to obtain the target's current position coordinate information (x RFID , Y RFID ). When the system is working, the multi-antenna radio frequency identification device may recognize multiple random tags at the same time. When multiple tags transmit signals to the multi-antenna radio frequency identification device at the same time, the signals interfere with each other, resulting in a signal identification conflict problem. In order to be able to accurately identify tags, a random number tree splitting anti-collision algorithm is used to solve the conflict problem. The specific method is: when multiple tags fall into a signal recognition conflict, the random number generator in the conflicting tag generates a random number 0 or 1. According to this random number, the conflict subset is divided into two subsets L (the random number is 0 Label set) and R (label set with random number 1). In the next transmission time slot, tags belonging to subset L will be sent. If there are still conflicts, subset L will continue to split. This process is repeated recursively until all tags are correctly identified;
[0044] Step 2. Image sensor activation and positioning: When the multi-antenna radio frequency identification device detects that the moving target enters the video surveillance area, it transmits an instruction to activate the image sensor. After receiving the instruction, the dormant image sensor turns into an active state, and tracks and Capture the current position information of the moving target (x CCDi , Y CCDi ) i = 1, 2; when the moving target leaves the video surveillance area, it transmits an instruction from the dormant image sensor, and the area image sensor enters the dormant state after receiving the instruction;
[0045] Step 3. Data fusion: used to fuse the position information of the moving target acquired by the multi-antenna radio frequency identification device with the position information of the moving target tracked and located by the image sensor, including the following sub-steps:
[0046] 3.1) First-level fusion: According to the position information of the moving target captured by the image sensor (x CCDi , Y CCDi ), i=1, 2, using multi-view collaborative data fusion to locate the moving target;
[0047] If the moving target is in area 2 (see image 3 ), and the current position information of the moving target is tracked and captured by two image sensors at the same time, the positioning result after the first level fusion is:
[0048]
[0049] Among them, α is the first-level data fusion weight, calculated according to the following formula:
[0050]
[0051] Where d 1 , D 2 They are the distances between the moving target measured by image sensor 1 and image sensor 2 and the projection point of the lens center on the roadway floor.
[0052] If the moving target is in area 1 or area 3 (see image 3 ), or when the moving target is only tracked by one image sensor and captures its current position information when it is in area 2, the positioning result of the image sensor is the first-level fusion positioning result:
[0053] (x CCD , Y CCD )=(x CCDi , Y CCDi ), i=1, 2
[0054] 3.2) Second-level fusion: Perform a second-level fusion with the position information obtained by the multi-antenna radio frequency identification described in step 1 according to the first-level fusion data obtained from video positioning to improve the accuracy of the measurement. The positioning results after the secondary fusion are:
[0055] (x,y)=β(x RFID , Y RFID )+(1-β)(x CCD , Y CCD )
[0056] Among them, β is the weight of the secondary data fusion, calculated according to the following formula:
[0057]
[0058] among them, with They are the measurement variances of multi-antenna radio frequency identification positioning and video positioning. Measurement variance of multi-antenna radio frequency identification positioning The position of a fixed target in the detection range can be measured n times to obtain its position coordinate information (x i , Y i ), i=1, 2...n, convert it to distance Then d i , I=1, 2...n to get the variance. Similarly, the variance of video positioning can be obtained;
[0059] Step 4. If the image sensor does not monitor the moving target and cannot sense its position information, the multi-antenna radio frequency identification device is used for identification and positioning, and the positioning result is:
[0060] (x,y)=(x RFID , Y RFID )
[0061] The mine mobile target carries or carries an active electronic tag with an ID number, which is used by the multi-antenna radio frequency identification device to detect and identify the mobile target.
[0062] Reference figure 2 , Is a schematic diagram of multi-antenna radio frequency identification positioning. Among them, the mine mobile target tracking and positioning method places a certain number of multi-antenna radio frequency identification devices in restricted areas such as underground roadway personnel entering and exiting the wellhead, mining face and other key areas, and turning roadway blind areas. These multi-antenna radio frequency identification devices are connected to a monitoring host on the ground through a data bus. When receiving wireless tag information, the data bus transmits the information in the wireless tag to the ground host, and the host locates and summarizes the information. Under normal circumstances, a multi-antenna radio frequency identification device is installed at each roadway opening, and then a certain device is arranged in the roadway according to the coverage of the multi-antenna radio frequency identification device to ensure that the entire roadway is covered. When the mobile target enters the underground tunnel area, the multi-antenna radio frequency identification device will detect the electronic tag carried by the mobile target, and use the RSSI-based triangulation algorithm to obtain the current position coordinate information of the mobile target (x RFID , Y RFID ).
[0063] Reference image 3 , Schematic diagram of CCD video multi-view fusion positioning. The specific control process of the multi-sensor radio frequency identification device to the image sensor is as follows:
[0064] When the multi-antenna radio frequency identification device does not detect a moving target, all image sensors are in a dormant state. According to the moving target position of the multi-antenna RFID device (x RFID , Y RFID ) And the known position of the image sensor device and the effective monitoring range of the image sensor. When the moving target enters area 1, it can be determined that the target enters the CCD 1 Monitoring area, wake up the CCD at this time 1 Perform image acquisition and processing.
[0065] When the moving target enters area 2, it can be determined that the target is in the CCD at the same time 1 And CCD 2 Monitoring area, wake up the CCD at this time 2 With CCD 1 At the same time, image acquisition and processing of moving targets;
[0066] When the moving target enters area 3, the target has left the CCD 1 Monitoring area, at this time, the CCD 1 Go to sleep.
[0067] Reference Figure 4 , CCD video positioning principle diagram. The positioning principle is described as follows:
[0068] The CCD device uses a small hole imaging model to describe the perspective transformation of the three-dimensional scene projected onto the two-dimensional image plane (CCD photosensitive matrix surface) of the CCD; where f, A, and h are the effective focal length, pitch angle and installation height of the CCD ( The height of the lens center to the ground), the intersection of the optical axis and the image plane is taken as the origin of the image plane coordinate system, generally taken as (0, 0); (x, y) is the projection coordinate of the lowest point p of the target on the image plane.
[0069] Using CCD to acquire video, the deflection angle of the target position relative to the CCD viewing axis can be calculated through calibration. Since the height of the CCD camera is known, the distance between the point P and the lens center on the ground projection point Q can be obtained according to the geometric relationship. The distance d between the personnel and Q point measured by CCD is:
[0070]
[0071] Since the CCD position information is known, the moving target position (x CCD , Y CCD ).
[0072] Reference Figure 5 , Is the block diagram of the multi-antenna radio frequency identification device. The device is composed of an antenna module, a microprocessor unit, a transceiver, a communication interface, a storage unit, and a power supply module, and is used for target recognition and positioning of underground tunnels. Among them, the antenna module is a 2×2MIMO antenna; the microprocessor unit is a component that realizes the communication protocol between the multi-antenna radio frequency identification device and the electronic tag, which can simultaneously complete the functions of receiving data signal decoding and data error correction; the transceiver includes The transmitter and receiver are two parts. The transmitter sends electromagnetic wave signals. The receiver is responsible for receiving the data signal returned by the tag and transmitting it to the microprocessor unit. The transceiver is connected to the antenna module; the communication interface is provided for the device and external entities. Communication instructions, through the controller to transmit data and receive instructions and respond; the storage unit is used to store the configuration parameters of the device and read the list of tags; the power module provides power for the entire device. The multi-antenna radio frequency identification device is an intrinsically safe explosion-proof equipment.
[0073] The MIMO antenna used in the present invention can effectively increase the coverage of the radio frequency identification device, and at the same time can improve the capacity and reliability of the channel; using the multi-antenna radio frequency identification positioning information to guide the opening or sleep of the multi-CCD image sensor can effectively reduce the downhole CCD image sensor Work energy consumption; at the same time, this method combines the advantages of rapid positioning of multi-antenna radio frequency identification and accurate positioning of CCD video, which can effectively improve the real-time and accuracy of tracking and positioning of moving targets in coal mines.
[0074] Obviously, those skilled in the art should understand that the above method steps involved in the present invention are not only used as a moving target tracking and positioning method in underground coal mine environments, but also suitable for non-metal and metal mines after proper integration or improvement. Mobile target monitoring, video tracking, pattern recognition or mobile positioning, so that the present invention does not limit mobile monitoring and target recognition other than mobile tracking and positioning.
[0075] The above content is a further detailed description of the present invention in combination with specific preferred embodiments. However, the protection scope of the present invention is not limited to this. Any person skilled in the art will not deviate from the premise of the design idea of the present invention. Any changes or substitutions that can be easily conceived shall all be considered to fall within the protection scope of the present invention.