摘要
为了解决经典图像异常辨识算法中图像辨识率不高、稳定性较差的问题,提出了一种改进算法。将先进的BP神经网络算法理论改进并引入到图像异常识别领域,用经过BP神经网络训练后的相关函数进行图像异常辨识,由于该算法充分考虑了图像像素的位置特征,并能根据图像内容进行自我学习,同传统采用灰度直方图进行辨识的算法相比,具有自适应、鲁棒性强的优点,可获得更高的辨识率。试验结果表明,同传统方法比较,该算法在稳定性和图像辨识率等方面都有明显提高。
To solve low efficient and bad stable problem of graphic distinguishment by classical graphic distinguish method, a new method is put forward in this paper, bringing an advanced method of BP neural network into graphic distinguishing area and using this related algorithm to distinguish graphic abnormality after BP neural network training. According to the content of image, this method can learn by itself, because of considering the position character of image pixels and compared with the traditional recognition method of using gray histogram, this method has fitness and strong robustness. The result of experiment shows that this method is obviously improved.
出处
《辽宁工程技术大学学报（自然科学版）》
EI
CAS
北大核心
2007年第5期740-743,共4页
Journal of Liaoning Technical University (Natural Science)
基金
辽宁省教育厅科技研究基金资助项目(05L169)
