摘要
针对车牌字符在车牌图象质量退化时识别率较低以及识别时间较长的问题,提出一种基于量子神经计算的车牌字符识别方法。该方法将通用量子门组作为神经网络的激活函数来实现量子神经计算,同时把字符的粗网格特征作为字符的识别特征进行车牌字符识别。实验结果表明,该方法能有效提高"带噪"车牌的识别率以及抗干扰能力。
Aiming at the problem of lower recognition rate and longer recognition time for license plate character when the license plate image quality degenerates, a license plate character recognition method based on quantum neural computation is proposed. This method implements quantum neural computation by making universal quantum gate as neural network＇s activation function, and chooses the rough grid feature of the plate as the character recognition feature to conduct license plate character recognition. Experimental results show this method can improve noisy license plate recognition rate and anti-interference ability effectively.
出处
《计算机工程》
CAS
CSCD
北大核心
2008年第23期227-229,共3页
Computer Engineering
基金
国家自然科学基金资助项目(60673092)
教育部科研重点基金资助项目(205059)
江苏省高校自然科学基金资助项目(07KJD520186)
关键词
量子神经网络
量子门
车牌字符识别
quantum neural network
quantum gate
license plate character recognition
