Research On Reading Recognition Technology Of The Gas Meter Based On Genetic Algorithm

With the development of science and the enhancement of computer technology, digitalimage processing has been extensively applied. As people’s living standards and theenvironmental protection consciousness have been gradually improved, the demand of naturalgas significantly increased. As an important management method, automatic meter readingsystem will be a future trend.The reading recognition of gas meter is the key process of the automatic meter readingsystem. The thesis studied the reading recognition of the gas meter based on genetic algorithm.We divide the procedure of recognition into preprocessing of images, positioning of images,tilt correction and reading recognition, and so on. The main research contents are as follows.In the stage of the image preprocessing, the collection image is transformed to gray scaleimage. Comparing three different filtering methods, the median filtering method is used forimage filtering operation. The Ostu method is used for binary image.In the reading area positioning stage, based on related position methods, the projectionalgorithm is used to achieve the coarse position of reading areas. Respectively, five kinds of edgedetection are used for image edge detection. By contrast, Canny edge detection operator is usedin this thesis, which can reduce the interruption of template image edge detection and get a betterbalance between edge detection and noise suppression. Canny edge detection and mathematicalmorphology algorithm are combined to achieve precise position of the reading area. Throughaccurate positioning, reading area is separated from the image of the gas meter, ready for therecognition stage.In the reading area tilt correctioning stage, Canny operator edge detection andHough transform are combined to achieve a good image In this thesis. The lens and the gas meterplane may be tilted, which has a great impact on the recognition rate, so it is necessary to tilt theimage for correction.In the reading recognition stage, the genetic algorithm is used to recognize reading images.The genetic algorithm is a method of searching the optimal solution. It is designed by simulatingthe natural evolution of Darwin’s theory and the theory of genetic variation. It owns a simpletheory and techniques, which is suitable for parallel processing features. The genetic algorithm isused to search the optimal solution, which startes from a population which may generate apotential solution and produces a new group of individuals better adapted to the environment bythe random selection, crossover and mutation operations. After evoluting to a better area,epigenetic generation population will obtain the better adaptability to the environment than theprevious generation. In this thesis, the known digital mtrix is used to search area image in theprecise location. It is not necessary to extract any feature when seeking the optimal solution. The experiments results show that reading recognition based on genetic algorithm can getgood results when we identify some different areas of gas meters, to, select a different initialconditions, such as the population size, the number of iterations, etc., and select a differentnumber of matrix and different regions of the image size. The recognition rate is more than80%,which testifies the effectiveness of the proposed recognition algorithm.