Abstract: Feature extraction and classifier design are the key link of handwritten circuit symbol recognition system.Because handwritten circuit symbol has irregular character，an approach to feature extraction that focuses on vision is proposed，on the basis，improved BP neural network of self-adaptive learning rate is used for recognition.By the experiment of classification in ten handwritten circuit symbols，it proves that the designed recognition system has very good result of classification and strong practical applicability.

Key words: handwritten circuit symbol, feature extraction, self-adaptive learning rate, BP neural network

摘要： 特征提取和分类器设计是手绘电路图形符号识别系统的关键环节。针对手绘图形不规则性的特点，提出了一种基于视觉的特征提取方法，并利用自适应学习速率的改进型BP神经网络进行分类识别。通过对10种手绘电路图形符号的分类实验，验证了文中设计的识别系统具有很好的分类效果和较强的实用性。

关键词: 手绘电路图形符号, 特征提取, 自适应学习速率, BP神经网络