Design of Infrared Moving Target Recognition and Tracking System Based on DSP+FPGA

The biggest feature of the DSP+FPGA structure is its flexible structure and strong versatility. It is suitable for modular design and can improve the efficiency of the algorithm. At the same time, its development cycle is short, the system is easy to maintain and expand, and it is suitable for real-time digital signal processing.

Compared with general-purpose integrated circuits, ASIC chips have the advantages of small size, light weight, low power consumption, and high reliability, and can reduce costs when used in large quantities. Field programmable gate array (FPGA) is developed on the basis of a dedicated ASIC, which overcomes the shortcomings that the dedicated ASIC is not flexible enough. Compared with other small and medium-sized integrated circuits, its main advantage is that it has a strong flexibility, that is, its internal specific logic functions can be configured according to needs, and it is convenient to modify and maintain the circuit. The biggest feature of the DSP+FPGA structure is its flexible structure and strong versatility. It is suitable for modular design and can improve the efficiency of the algorithm. At the same time, its development cycle is short, the system is easy to maintain and expand, and it is suitable for real-time digital signal processing. This article introduces the design of an infrared moving target recognition and tracking system that can be applied to military reconnaissance.

Design tasks and requirements The input signal of the infrared moving target tracking and recognition system is an analog or digital video signal provided by an infrared camera. The system uses a digital video processing card based on the C6X series high-speed DSP to process infrared digital video sequences in real time to complete the search, capture, tracking, and memory of moving targets; and displays infrared video images on a PC in real time to give moving targets in real time The spatial coordinates of the target are generated, and the characteristic data of the moving target area is generated, and the real-time storage or remote transmission of the image of the moving target area is completed. The hardware module needs to provide hardware support for the realization of the system function, that is, to provide the underlying physical support that is compatible with the system function, including operation processing speed and storage capacity.

Analog video digitization precision requirements: AD precision is 8bit; digital video channel requirements: receive data according to RS422 transmission protocol, pixel precision 14bit; image processing time per field 40ms; search capture time: 0.2 ~ 1s; capture tracking time 120ms; 25 frames/s real-time identification and tracking of moving targets (that is, the current field data must be processed before the next field data arrives, and the computer outputs the processing results and displays the video image), and gives the target position and field image; interface with the computer It is a PCI interface.

UNIT 1 module is based on the standard 32-bit +5V PCI bus, and is equipped with a very large-scale programmable chip (DSP, FPGA), which has strong computing and processing capabilities.

The function of UNIT 2 module is mainly to realize the detection and tracking of moving targets under the moving background. Considering the real-time requirements of the system, the moving target detection in the moving background uses a differential technique based on camera motion compensation. First, the global motion caused by the camera motion is compensated, and the compensated sequence image is differentially calculated; then the motion disturbance region caused by the target motion is searched in the differential domain; and finally, the motion target is extracted on the original video image. At the same time, the prediction technology is used to estimate the possible location and existence area of the target, so as to achieve real-time and accurate tracking (or memory) of the target. The system software is divided into four status modules according to its working status: search, capture, tracking, and memory tracking. The system operates in accordance with the four states of search, capture, tracking, and memory tracking and their conversion to achieve real-time detection and tracking of moving targets.

The main function of UNIT 3 module is to realize the data communication and information interaction between the hardware module and the upper application. The system uses the single-cycle read/write of the PCI 9054 Target method; in order to meet the real-time transmission and processing requirements of 25 frames per second of image data transmission, the PCI 9054 Scatter/Gather DMA method of data transmission is used. In the information exchange of the entire system, a one-time handshake protocol is adopted, that is, a request-to-one response protocol.

The main functions of the UNIT 4 module are to download the DSP tracking program to the hardware module, start/stop the DSP, display the scene video in real time, store the moving target sequence in real time, and analyze the basic characteristics of the moving target sequence in real time and display the results.

For the convenience of design and description, we divide the circuit structure of the hardware module into the following units: video interface unit, input and output FIFO video image memory, digital image processing unit ( DSP), programmable controller, PCI interface circuit with PC, etc.

1 Video interface unit

The video source of the infrared moving target recognition and tracking system is the video signal provided by the infrared camera. The infrared camera has two video outputs, namely analog video output and digital video output. This system requires hardware modules to be able to process both video signals. Therefore, the input video signal must be preprocessed to provide the necessary video data and video synchronization data for the digital image processing unit (DSP). 

2 input and output buffer FIFO

The purpose of setting the input and output buffer FIFO is to set up a buffer between the high-speed device and the low-speed device, which can avoid the high-speed device waiting for the data of the low-speed device to reduce the efficiency of the system. The clock frequency of the digital signal sent by the A/D chip is about 12MHz (analog channel clock 12.51MHz, digital channel clock 12MHz), and the bus frequency of the DSP on the processing card is up to 50MHz, the difference between the two is large, so the input and output buffer FIFO is used necessary. Based on the above considerations, Cypress's CY7C4275 was finally selected. Its capacity is 32K×18, and the maximum access speed can reach l0ns.

3 programmable controller (FPGA)

In this system, FPGA controls most of the units, including channel selection/level conversion chip, input and output FIFO, SRAM, DSP, PCI interface circuit, etc. Use the in-system programmable (ISP) performance of the FPGA chip to complete the control logic of all DSP peripheral chips; and set the status register, command word register and special register in it to complete the real-time communication with the host and receive the command information and Transmit the required status information to the host.

In this system, the digital channel is 14bit and the simulation is 8bit. The FPGA needs to assemble the signal for the first time (different from the second time that the DSP assembles the image in FGB format for display), that is, the digital/analog signal/ The data are converted into 16bit data, and then two 16bit data are assembled into one 32bit data.

4 Digital Image Memory (SRAM)

The infrared moving target recognition and tracking system should complete the recognition and tracking of moving targets. Its implementation algorithm must involve the processing of multi-frame (differential processing, at least two frames) video images. In order to provide sufficient storage space for the implementation algorithm, the memory we selected can accommodate 6 field video images. Therefore, the last selected memory is Giga Semiconductor’s two GS74116s, each with a capacity of 256K×16bit and an access speed of 15ns. Considering that the data volume of each field of our video image is 76800 pixels, two 512K SRAM can Save at least 6 video images. 

DSP uses TI's TMS320C6202 chip. We use sub-sampling with interlaced and interlaced lines. After sampling, the image size of each frame is about 20KB, and a total of about 80KB of data space is required. The on-chip data space of the TMS320C6202 is sufficient. 

function ImgZoom(Id)//Re-set the image size to prevent breaking the table {var w = $(Id). width; var m = 650; if(w 6 PCI interface circuit Because the interface between this system and the PC is a PCI interface. In order to avoid being trapped by the complicated data transfer protocol of the PCI interface, to fully utilize the data transfer capability of the PCI bus, the PCI interface circuit Adopt PCI9054 chip. It is a dedicated interface chip for PCI bus, which has the advantages of fast data transmission and simple data transmission. At the 33MHz PCI bus operating frequency, its maximum data throughput capacity is 132MB/s. PCI9054 and DSP data exchange or Communication is achieved through two registers inside the DSP chip, namely the address register XBISA; the data register XBD. 

The system software design first conducts a system power-on self-check, checks whether each part of the system enters the normal working state, and sends the test result to the host. Then initialize the entire system, check the command word register to determine the image input mode and the system's working mode, if the host is not specified, enter the wait state, until the operator specifies the system's working mode, the system enters normal work.