摘要
提出了一种基于协同学理论的不变性模式识别模型．该模型首先对输入图象模式进行位置和大小上的归一化，然后提取其旋转不变性子波特征作为模型的试验模式，最后由协同系统的序参量迭代方程得出正确的识别结果．这种方法与Ｈａｋｅｎ提出的两种协同不变性模式识别方法相比，既比频域法更接近于人类的认识过程，又避免了“伪状态”的出现，使得系统严格收敛到正确的模式上．实验表明，该模型不但具有很好的识别精度，而且还表现出较好的抗干扰和缺损能力．
A PSRI(position, rotation, scale invariant) pattern recognition method is proposed, which does the normalization of the position and scale of the input unknown image patterns and extracts the rotation invariant features using radius wavelet transformation as the initial system state vectors and finally the correct results are acquired from the iteration models of the order parameters of synergetic system. Compared with the two PSRI pattern recognition methods proposed by Haken using synergetic theory, this method is more similar to the thinking process of human and avoids the occurrence of false states as well. The experiments show that the proposed method achieves the satisfactory recognition results on the condition of noise and deformity.
出处
《上海交通大学学报》
EI
CAS
CSCD
北大核心
1998年第6期1-3,共3页
Journal of Shanghai Jiaotong University
基金
国防预研基金
国家自然科学基金
关键词
不变性模式识别
协同
子波变换
归一化
PSRI pattern recognition
synergetic
wavelet transformation
normalization
