Research Of The Energy Meter Detection System Based On Machine Vision

In order to improve the factory quality and the production efficiency of the energy meter further, and reduce costs and labor intensity, it is of important significance to apply the machine vision to the energy meter detection. This paper shows that this study is on smart meters. In this paper, firstly, the technology of energy meter detection system is researched. Secondly, its hardware environment is designed reasonably. Thirdly, its image processing algorithms are studied. In the final, the software of energy meter detection system based on machine vision is developed. Then, this paper shows the image processing, detection and recognition of the energy meter, and it demonstrates the system has visualization. Through the theoretical analysis, the experimental results show that the energy meter detection system based on machine vision can complete energy meter detection. This main work in this paper is as follows:As a whole, this paper designed power meter detection system based on machine vision, and what’s more, it shows the structure and workflow of the system about the hardware environment and software design. In terms of hardware environment, the hardware structure of detection system is researched, the influence of each component on the system is introduced, and then, our work focuses on comparing the influence of the various lighting systems on obtaining high quality images of energy meter; in terms of software design, software development environment and the design process are introduced.In terms of image correction, this paper introduces rotation correction, image redundancy reduction and normalization. In allusion to the problem about floating point coordinates in the correction, the optimum interpolation algorithm is proposed to suit this system. According to rectangle features of LCD on energy meters, it is proposed that the contour detection combined with polygonal approximation is used to locate LCD. Based on the analysis of existing barcode location methods, and the relative positional relationship between each of the feature region, it is successful to locate the barcode area.The characteristics of the character region are studied. In the next, it is proposed to use horizontal integral projection combined with vertical integral projection for segmentation of a single character, and use the template matching method to identify characters. Meanwhile, the barcode encoding rules of the smart meter is studied. It is better to apply the interlaced scanning combined with local computation in barcode decoding.This paper analyses the type of energy meter defects, uses template matching and image difference for detecting defects in the feature area and apply Blob analysis for the defect difference image. At the same time, our work also focuses on the calibration for the defect area of the energy meter, identifies the type of defect, and studies how to discriminate the type of defects. The validity of the theory is verified in the experiments. At last, through statistics on all the information of the energy meter, it is easy to inquire.