Research On Automatic Identification Technology Of Pointer Meter Reading Based On Image Processing

Pointer instrument has many advantages, thus there is a huge stock of production which is increasing every year. Because pointer instruments need a lot of attention of checking, and artificial reading is not only high efficiency and low cost, but also reading error happens from time to time. As digital image processing technology has gradually matured, its application to the automatic identification system pointer instrument is an effective solution.At present there is lot of research in this area, and some automatic identification systems are designed. These systems have research deep into hardware feasibility, and about recognition algorithm, they focused on extracting the pointer straight normally, which under various constraints can achieve a bit more accurate recognition results. However, meter is working under various kinds of environment, moreover, there is a wide range of instruments, constraints severely restrict the universal of automatic identification system pointer instrument, and it is difficult to ensure the identification system to maintain high accuracy readings in transplantation.This paper has research into instrument pointer automatic identification system which applied domestic and oversea, especially a solution based on image processing is put forward considering the shortcomings and limitations of existing automatic identification system and. The main work and innovations are as follows:(1) Melioration of the image processing algorithm. There are different designs for Image processing in different areas, contrapose the problem of noise, light and shadow when pointer instrument recognition under operation, an image processing algorithm is present which applied in automatic identification system, thus anti-jamming capability of recognition system is improved.(2) Dial extraction algorithm is proposed. Pure dial image recognition is the starting point for almost all existing systems, while the image actually operated under complex background. Dial extraction algorithm detects the edge meter image firstly, then a different algorithm extracting for square and round dial. On the square dial, straight line detection is applied, and based on statistical algorithms use a straight line to extract dial border characteristics; for circle detection algorithm use circular dial to extract the dial.(3) Improved pointer feature extraction method. Panel meter is varied, which bring with the diversity of pointer and dial the environment, automatic identification system should be able to extract the correct features under a variety of situations. In this paper, the skeleton extraction takes place of edge features, and to reduce pointer feature extraction interference, bur is applied, which solve the identification system's portability problem.(4) Specification of pointer reading's calculation method. Existing studies have not standardized and universal pointer readings computing model, this paper based on the idea of the angle method, a new model of reading pointer is proposed, using two images to obtain a priori knowledge, reducing human intervention, which improves the reading pointer precision, and more importantly, can apply to all types of pointer instruments.