基于神经网络的图像识别

Image recognition based on artificial neural network

神经网络图像识别技术是随着当代计算机技术、图像处理、人工智能、模式识别理论等发展起来的一种新型图像识别技术。在进行图像识别之前需要利用数字图像处理技术进行图像预处理以及特征提取。本文选取字符图像0—9作为识别目标，对图像预处理过程进行了叙述，并在此基础上选取字符图像矩阵每行的与每列的黑色像素点之和以及图像欧拉数这两个特征作为BP神经网络的输入样本。经实验仿真表明图像的平均识别率为89％，这表明图像预处理的结果和提取的特征是合适的、有效的。设计的BP网络也较好的完成了模式分类识别工作。

Image recognition based on artificial neural network is a new type of image recognition technology that develops with the development of modern computer technology, image processing, artificial intelligence, pattern recognition theory and so on. It is necessary to make the image pre-processing as well as the feature extraction work using the digital image processing technology before carrying on the image recognition. In this paper, the character image from 0 to 9 is chosen to expound the process of image pre-processing and based on this, the sum of black pixels of each row and column and the Euler numbers are extracted as the BP neural network input sample. The results of experiment, average recognition rate of 89%, can prove that the results of the character image pre-processing and extraction of features are appropriate and effective and the designed BP network is also better to complete the pattern classification and recognition.