The Research On Instrument Recognition Technology Based On Machine Vision

Machine vision technology through computer simulation of the human eye, based on the extracted image information,on the object and scene forms and modes of motion can be extracted by image information to judge,and finally applied to the actual production and life. In the production process,application of a pointer type is very common.When conducting scientific exploration of people to use the instrument, most production practices should be used in the process instrumentation, instrument for the people through measurement and control, the large number of instruments, many species are, people can Instrument obtain all kinds of information, it is apparent objective world of human knowledge can not be separated instruments. Industrial instrumentation reflects the level of technological development of a country, a country which is the source of the information obtained. Currently, the main pointer instrument readings are reading through the manual, manual labor because of repeated readings, high labor intensity, inefficient, error-prone, it is necessary to use machine vision technology for automatic identification of the instrument.This paper is the research on instrument recognition technology based on machine vision, focusing on the pointer instrument image preprocessing methods, feature extraction method pointer, and to identify the meter readings. First, starting from the idea of machine vision, combined with the basic method of digital image processing, for pointer instrument gray-scale image processing, due to the pointer instrument collection process may be affected by changes in illumination, you can use homomorphic filtering method image enhancement processing to improve image quality, lay the foundation for recognition behind, homomorphic filtering parameter selection is difficult, often using the method of trial and error, time-consuming and there is no universal, we propose a homomorphic filter parameter selection method, and then the image erosion and dilation, image binarization, thinning and other pretreatment. Then, line extraction method introduced, and extract the pointers to give the characteristic parameter pointer with Hough transform. Finally, with the point scale online fitting center coordinates, ie, the rotation center coordinates pointers, while using angle method to calculate the indicator reading, and finally with Matlab simulation, and error analysis.