Research On High-speed Image Acquisition And Preprocessing Technology Based On FPGA

The key of hot rolling slab surface temperature measurement system based on radiation thermometry techniques is to acquire high-speed, accurate slab surface images. However, it is difficult for the traditional image acquisition and processing system based on PC to achieve both high-speed and high-accuracy. In this paper, we proposes a high-speed image acquisition and preprocessing system based on FPGA. This system puts the wavelet threshold de-noising processing algorithms based on the hot mill site environment forward and operates them with FPGA. These complex image processing algorithms are able to ensure accuracy without affecting the image acquisition and processing speed.This system adopts a master-slave FPGA architecture using PCI as transmission interface to connect with PC. This structure and interface ensure high-speed image acquisition, processing and the feasibility of transmission and make it available to modify the image preprocessing algorithm flexibly for the grabber according to the request. In this paper we describe the hardware device selection, circuit schematics and the PCB map design about the design scheme of the overall system. In addition, this paper provides a detailed debug step of hardware circuit debugging and the summary of the process encountered.The software part of the system is divided into four parts, which are:FPGA control module, PCI drivers, Windows applications and image preprocessing algorithm module. In this paper, FPGA control module introduces the video decoding module, the programming method of SDRAM control module and PCI interface module in detail; the design of the PCI drivers and Windows applications achieve the communication between acquisition card and the host computer. The image preprocessing algorithm modules describe the FPGA-based 5/3 lifting wavelet threshold de-noising algorithm design in detail. When realising the 5/3 lifting wavelet decomposition algorithm, we propose a method, which is called module reuse method, to save resources. The paper also proposes a transformation-based discrete-time system modeling method, which can reduce the difficulty of designing the system greatly.Finally, we verify and summarize the design of image acquisition and processing system, obtain detailed validation data and talk about the advantages and disadvantages of this system.