Research Of The Gauge Wheel Number Recognition Algorithm Based On Image

The gauge wheel meter is widely used in household use and industrial areas, but at present the household meter calibration device has low degree of automation, and its data acquisition work still needs a lot of artificial operation, which has a negative effect on work efficiency. Except the commonly artificial reading, some existing reading methods such as photoelectric sampling method are all insufficient in the process of application. Therefore, designing a more effective recognition method for the gauge wheel meter has a certain practical value to improve the efficiency of industrial production and testing.The research on the number recognition of the gauge wheel meter, not only needs to design various character recognition algorithms under the ideal situation, but also needs to consider the impact of the ambient light interference, skew images and different character sizes on the result of recognition. According to the above problems, the main research content of this paper is as follows:1. In the process of the image skew correction, aiming at the traditional Canny algorithm has poor effect in extracting the image edge and weak dynamic performance, an improved Canny edge detection algorithm was studied. On the basis of the traditional Canny algorithm, the method used the Otsu algorithm to dynamically determine the high and low thresholds to detect the edge according to the gradient properties of the image itself, which can preserve the image edge details, at the same time, suppress a certain amount of noise interference, has strong self-adaptive ability and improves the precision of the straight line detection.2. Because the impact of the border of the character leads to the difference between the scale and the character part, using a single image segmentation method can’t have a complete split of both scale and character. To solve this problem a method by combining two segmentation methods was adopted, first using global threshold segmentation method to make an image binaryzation, after locating the scale using the local threshold segmentation on the gray-scale image, which can make full segmentation of scale parts in the image. Experimental results showed that the method can completely split out of scale and character parts on the basis of keeping the faster speed.3. A template matching method based on the geometric feature and shape feature of the characters was designed to recognize the meter number. The image fragmental processing to the templates and the images to be identified was made, and on the basis of character image normalization, a statistics about the center of mass and area of the target in each section was made, and using the least squares theory can find template which is the most similar to the character to be identified, then the character corresponding to the template is the requested character. After experimental verification, the method on the complete character and half character recognition has high successful rate.With experimental verification and analysis, the method in this paper can achieve function in the gauge meter number recognition and has good recognition effect.