Pointer instrument representation value identification system based on wireless sensor network and convolutional neural network

Abstract

The invention discloses a pointer instrument representation value identification system based on wireless sensor networks (WSNs) and a convolutional neural network (CNN), and relates to the technical field of wireless sensor networks, image processing and convolutional neural networks. The system is composed of a WSNs terminal node, a WSNs coordinator node and upper computer software. According to a recognition algorithm, pointer rod coordinates are extracted through image preprocessing, a structure-optimized CNN model (the pointer rod coordinates serve as model input) is used for classification to obtain a large scale value, and then an angle method is used to calculate an indicating value. According to the system, parameters (the number of large scales and the measuring range of the instrument) of a to-be-recognized instrument are set through an upper computer, image collection and algorithm recognition of a pointer meter are completed on the WSNs terminal node by means of the data processing capacity and the communication capacity of the WSNs terminal node, a recognition result is sent to a coordinator node through a wireless network created by the WSNs coordinator node, and a coordinator uploads the received indication result to the upper computer for display through a USB. The system node can be directly powered by a battery, and can be quickly and conveniently arranged in a temporary important place to complete the identification of the indication value of the pointer type instrument.

Description

technical field [0001] The present invention relates to the technical fields of wireless sensor networks, image processing and convolutional neural networks, in particular to a method for recognizing pointer-type instrument representation values ​​based on wireless sensor networks (WSNs) and convolutional neural networks (CNNs). Background technique [0002] Although smart instruments have been widely used in various fields, pointer instruments without digital output interfaces have inherent advantages, such as low cost, high reliability, and simple structure, and are still used in modern industrial processes to monitor different equipment and process parameters. The indications of pointer instruments are usually obtained manually by technicians, which is inefficient. In order to solve this problem, the automatic reading method of pointer instrument based on computer vision has been widely researched and applied. However, most of the existing pointer instrument identificat...

AI Technical Summary

Problems solved by technology

However, most of the existing pointer instrument identification systems are wired systems. The disadvantages of wired systems include: cumbersome installation, high maintenance costs, and lack of flexibility.

WSNs node resources are limited, and the algorithms running on them are usually small, while the parameters of the convolutional neural network model are huge.

Images