The Research And Design Of A Verification System About Pointer Water-Meter Based On Image Processing

Pointer water-meter is still the main measuring tool on water consumption, especially to household water-consumption, we are all basically use water-meter to have a test. According to the regulations of the state, water-meter manufacturers and metrological supervision and inspection departments need to detect water-meter accuracy according to the batch or on a regular basis. For most of the pointer water-meter, commissioners analyze the difference between scale reading and the afferent standard information by visual inspection. So long-time pointer interpretation causes verifier inevitably to appear the phenomenon of visual fatigue which makes great error in the test results or require a second interpretation about pointer readings when we check water meter is qualified or not.Pointer meter recognition system based on image processing is based on the computer, by making the system to see water meter dial image and transmit it to the computer. Computer automatically processing image and identify pointer meter readings. Thus the system can help or even replace the commissioner for water-meter calibration.This topic regards realizing the pointer meter’s interpretation automatically and improving the identification precision of the verification system as the goal. Based on further study of a large number of pointer meter identification system related image processing method, this paper put forward a way which combines with the feature of water-meter and makes full use of the water-meter image’s structural information to identify water-meter readings. Compared with other pointer instrument recognition system, the recognition system’s advantage is mainly reflected in:firstly, on the object, the system applicable scope is so broad that any red basal pointer can realize the extraction of pointer. Extraction uses the pointer the model whose name is "Luminance and red color difference" processing image and improving the discriminating accuracy about pointer-image. Secondly, for the determination of the direction of the pointer, this paper abandons the traditional methods of model matching and adopts the way about seeking the centroid of needlepoint’s pixels. Thirdly, this paper improves angle method to realize the improvement about the decipherable accuracy.Then this paper regards the accuracy about qualified rate of checked the instrument whose benchmark is indication error in different flow point of the system as main verify indicators. The experimental results, compared with other traditional methods, shows that the article using the image processing and recognition method has a good reliability.