Design Of Pointer Type Instrument Recognition System Based On ADSP-BF609

One way of digital pointer type instrument is to read and recognize the image of the instrument by image acquisition and processing technology, which can improve the accuracy and efficiency of the interpretation, and avoid the error caused by subjective reasons. In this paper, the digital technology of pointer type instrument based on image processing is studied, and the implementation scheme of the system based on DSP is presented.First of all, the paper studies the image format conversion, image denoising, edge detection, image binarization, Hough transform and other pointer instrument identification algorithm, and the simulation was done by MATLAB. By comparing the simulation results, determine the use of median filtering for image denoising, Canny operator for edge detection, Hough transform for central location and line detection.Using the ADSP-BF609 evaluation board as hardware platform, the control registers of the camera OV9653 are initialized by the TWI interface of BF609 to fufill image quality control. Through configure the parameters of EPPI1 interface and its special DMA controller accomplish the image acquisition and storage. The image format conversion of YUV to RGB is realized by configuring the parameter of the PIXC processing unit, which is used for LCD display. The image processing line is composed by the CNV module, the PMA module and the PEC module of PVP(Pipelined Vision Processor) to realize Canny edge detection operator and use the operator to detect the edge of gray image to reduce the computation of Hough transform. The display of the image is realized by EPPI0 port connection LCD. And through the Ethernet port for the transmission of data to the host computer.This paper is based on CCES integrated development environment to complete the software development and system debugging. The driver of OV9653 camera, PIXC, PVP and other hardware module is written according to the specific hardware features. Software porting the algorithms of image gray processing, median filter, Hough transform to locate the center of the circle, the image block, Location of the pointer to the partition and pointer angle calculation. After testing, by using PIXC hardware implement image format conversion is 10 times faster than Software implementation, and using PVP hardware to achieve Canny edge detection is 80 times faster than Software implementation and the extracted image edge information is clear. The calculation of the pointer type meter is accurate.In summary, pointer image recognition based on DSP has high accuracy and real-time. In the case of defaced dial,further study the algorithm, improve the applicability.