Automatic Recognition Research Of The Analog Display Instrument Based On The Figure Recognition

Recently, with the higher request for the automation level and the intelligence degree in the electrical power system, the computer vision technology has been gradually introduced in the automation of the electrical power system, and starts to play the increasingly vital role. The birth and the application of the computer vision technology has greatly liberated human labor power, raised the productive automation level, and improved the situation of human life, so its application prospect was extremely broad. This paper mainly discusses the computer vision and the correlation technology, proposes and has realized an automatic reading system of analog display instruments based on feature recognition.In the electric power industry, the analog display instrument is still the main measuring tool. It is still widespread applied in the industrial production and the scientific experiment process, and is playing the pivotal role. At present, the reading of analog display instruments is usually completed by the artificial reading, however this method is influenced by the human subjective factor, so the reading error of the data is very big, and the reliability is low. The purpose of this paper is to use the artificial intelligence method by computer to research the automatic reading system of analog display instruments in order to replace this arid and the tired work of artificially reading. We carefully analyses the artificial reading process. Firstly find the dial plate position according to the dial plate characteristic, then seek the indicator position in the dial plate region, observe the indicator and the closest two scale division line distance as well as its both sides numeral, make the reading of the mater. The reading based on the computer vision similarly experiences these three processes: the dial plate localization, the indicator localization and the numeral localization. Measuring appliance image is gathered by using the camera, according to the outline track method based on feature recognition realization the dial plate localization and indicator, digital localization, using two fitting least squares methods determined the parameter of the indicator, simultaneously using the BP neural network to carry on the recognition to the indicator both sides numeral, finally having realized the analog display instrument according to the indicator and the digital relative position relations.This article has analyzed the automatic reading algorithm of the analog display instrument based on feature recognition in detail, developed the platform using Visual Studio.NET, and used Visual C # to develop the automatic reading system of analog display instruments. Through carrying on the performance testing to this automatic reading system of analog display instruments, and carrying on the experimental contrast with the traditional Hough transformation examination law, the results finally show the characteristic extraction and the neural network recognition method which regarding the image has the distortion situation still have the high recognition rate, reading accurate, serviceable. Hence, this system has basically satisfied the request of the practical work, and has a certain value and prospects for development.