基于BP神经网络仪器显示自动识别方法

Automatic Recognition Method for Instrument Display Based on BP Neural Network

介绍一种基于BP神经网络仪器显示自动识别方法,是在VC++编程环境下实现的。仪器显示图像预处理主要包括倾斜度调整、图像去噪、特征提取。采用Hough算法调整图像的倾斜度,采用一些特定的模板对图像进行降噪,采用投影法提取图像的特征。实验表明这种方法运行速度快、识别率高。这种方法具有一定的实用价值。

The paper introduces an automatic BP neural network approach for instrument display recognition, which is realized in Visual C＋＋ compiling environment. Pretreatment of instrument display image mainly includes slope angle adjustment, noise elimination, feature distilling. The Hough transform is used to detect and correct the skew angle of digital instrumental image. Some special matrix templates is used to eliminate noise. One image projection method is used to extract the features of seven-segment numbers images. The experiment shows that the approach is a fast and high accuracy way of digital recognition. Therefore, the approach is feasible for use in seven-segment numbers recognition, in practice.