基于分形维数和隐马尔科夫特征的车牌识别

License plate recognition based on fractal and hidden Markov feature

针对现有车牌识别方法中车牌二值化和车牌字符识别效率不高的问题,提出一种基于分形维数和隐马尔科夫特征的车牌识别算法。该方法基于分形维数和隐马尔科夫特征并利用联合OC_SVM和MC_SVM的方法进行车牌识别。实验中,基于分形维数进行车牌的二值化处理;利用隐马尔科夫特征办法进行字符特征提取,然后利用多重分类器进行字符识别。对字符、英文字母和阿拉伯数字分别进行了800幅、800幅和1600幅图像的识别,得到的结果显示该算法对字符、英文字母和阿拉伯数字的识别率分别为98%、98.5%和98.9%,对各种不同的车牌整体识别的平均识别率高于90.60%。该方法识别效率高、鲁棒性强,为车牌识别的准确性提供了保证。

Because existing license plate recognition algorithm has lower efficiency in the binarization and character recognition,a license plate recognition algorithm based on fractal dimension and hidden Markov features was proposed.The algorithm is based on fractal dimension and hidden Markov features,and uses the joint classification of OC_SVM and MC_SVM to recognize license plates.In experiments,the fractal dimension was used complement the binarization of the license plate,the hidden Markov features were taken to extract character features and a multi-classifier was utilized to recognize the character.800 Chinese character images,800 English letter images and 1600 Arabic numeral images were recognized,obtained results show that the recognition rates of Chinese characters,English letters and Arabic numerals are 98％,98.5％ and 98.9％,respectively,the average recognition rate of license plates is more than 90.60％.It concludes that the method has higher efficiency,better accuracy and stronger robustness,and it can provide a guarantee for license plate recognition.