| Pointer instrument's structure, which is simple, cheap and easy to use, is still widely used in industry, agriculture and scientific experiments. However, the automatic identification and restrument is still a serious problem to be solved. It can realize the automation of the induadings of instries related to pointer instrument and raise the efficience of work and production.This paper mainly discusses the application of machine vision and image processing technology to the pointer instrument identification. Based on machine vision technology an automatic identification system is designed and realized. The system collect images by the CCD camera system, and transfer data through the USB cable to the computer for processing. System's core is the image processing algorithm, which choose and process the key frame according to the collected image's feature. After transforming color space the different leyers of the image are choosed and processed under different illumination. According to the panel's characteristics of pointer instrument, the image is preprocessed first, and then the template image is obtained by morphological processing to reduce image information redundancy. Hough transform and cubic spline interpolation are used respectively to locate the meter's pointer and panel's indicator. Finally the indication of pointer is obtained through the location result and simple algebraic transformation.Experimental results show that the system algorithm is simple, stable and has small identification error. For the 85C1 voltage meter with accuracy 2.5, the average relative error of identification can reach 0.29%, and system response time is also less than 0.2S so that it can meet the real-time requirements. When out of range it can automatically alarm and can better identify the pointer's image under different illumination. The system basically meets the requirements of practical work and application so that it has some application value and development prospects. |
PDF Full Text Request