Font Size: a A A

The Researching On Embedded Recognition Method Of The Power Pointer Meter Image

Posted on:2014-05-20Degree:MasterType:Thesis
Country:ChinaCandidate:G NiFull Text:PDF
GTID:2268330401966971Subject:Control engineering
Abstract/Summary:PDF Full Text Request
This paper takes the substation renovation subprojects as the applicationbackground. Contrast to the general image processing method and to the pointer’spositioning algorithm for pointer meter, this paper proposes a point meter recognitionalgorithm and implements an embedded system to verify it.This paper proposed a pointer localization algorithm which uses dial radius and therotation center as priori features based on Canny edge operator and Hough linedetection in this paper, in order to overcome the problem that the localization of thepoint was easily to be disturbed in the image recognition method. Firstly, to reduce theCanny operating time, the algorithm extracts the square region of interest by the centerof rotation center and sides of dial diameter. Secondly, it compares the meter’sstructural characteristics with straight after Hough transform, and then removes theinterference linear that does not satisfy the conditions of distance and position. Finally,it analyze and correct the error of value calculation combined with features of Cannyedge.Considering that the project’s application background and the principles ofbuilding embedded platform, this paper chooses an embedded Linux system withS3C6410microprocessor as the verification platform, the system which compared withcore hardware and programming language, and gives a brief introduction to thetransplantation of QT and OpenCV. The algorithm has an excellent performance of therecognition in all kinds of interference conditions on the platform.In the end, this paper concluded some shortcomings of this recognition algorithm,and then proposed some methods to improve it. Those methods include that adding agyroscope sensor to correct the error caused by image acquisition deviation, usingartificial neural network which adjust input parameters of Canny edge function toenhance the adaptability of the recognition algorithm to fit multi-sensor data. Then, thispaper summarizes the changes of hardware platform and programming language inembedded system by taking low-end electronic building blocks Arduino and high-endmultimedia application Raspberry Pi as examples.
Keywords/Search Tags:pointer meter, embedded system, Canny, Hough, image processing
PDF Full Text Request
Related items