Research And Implementation Of Pointer Instrument Reading Recognition Algorithm Based On ARM

The wide application of the pointer instrument in the production practice makes the application of image processing technology to monitor it become a hot research content.For the instruments used in power system scenarios,this paper combines the image processing technology and embedded technology,studied the reading recognition algorithm for the pointer instrument,and selects the embedded development board with microprocessor S3C6410 as the core,based on Qt/Embedded A graphical user application program was written,which realized the accurate identification of the pointer instrument reading under the embedded platform.This paper elaborates the principle of ARM-based pointer meter reading recognition algorithm,analyzes the process of reading recognition based on image processing,and completes the algorithm migration under embedded platform.The main work includes:(1)Analyzed the research background and significance of the subject,studied the literature data related to the finger-type meter and the liquid level meter reading recognition in embedded platforms in recent years,and deeply understand the use of embedded development environment.(2)The recognition algorithm based on projection and angle is used to recognize the pointer instrument readings.The Otsu method(maximum inter-class variance method)is used to extract the pointer area.According to the mathematical relationship between the rotation angle and the instrument scale projected on the vertical axis of the pointer area,the Otsu method is used to identify the readings.Through the Hough line transform to extract the level,according to the center of gravity position and the ranks of pixel scanning to identify the number of scales,and ultimately the realization of the liquid level meter readings.(3)The Feilink OK6410 development board with microprocessor S3C6410 as its core is selected to build a hardware platform and program the embedded Linux operating system.(4)The graphical user application program based on Qt/Embedded is written,and the image processing function is added.By using the cross-compiling environment,the embedded transplantation of the reading recognition algorithm is realized.The Qt/E application program to open the image on the development board,save,enlarge,reduce,suitable for screen operations,as well as image grayscale,image filtering,tilt correction,feature extraction,feature recognition and other functions.(5)Acquisition of the pointer instrument and liquid level image at the site of Baoding Thermal Power Plant,and completed the application test of reading identification based on ARM.The test results show that the recognition error is less than the error of manual estimation,and the correct recognition rate of the readingresult is more than 98%,and the average recognition time is less than 1s,which can meet the accuracy and real-time requirements of instrument reading identification under embedded platform.