A novel intelligent classifier/recognizer employing vector quantization coding of nonorthogonal signal and preprocessed signal representations

Classification and recognition of one and multidimensional signals is one of the important tasks in emerging technologies, particularly communication systems. In this contribution, novel digital signal processing techniques are employed to achieve this task. Classification decision tree algorithms have been used recently in pattern recognition problems. In this paper, we are proposing a self-designing system, using the classification tree algorithms, capable of recognizing a large number of signals. Preprocessing techniques are used to make the recognition process more effective. A combination of the original as well as the preprocessed signals is projected into different transform domains. A large number of criteria, that characterize the signals, can be developed in these domains. At each node of the classification tree, an appropriately selected less-complex and more noise-immune criterion is optimized and employed by the vector quantization (VQ) design techniques to divide the signals, presented at that node in that stage, into two approximately equal groups. The use of VQ leads to high classification accuracy even with noisy data. At the terminal node of the tree, each signal is represented by a unique composite binary word index. Experimental results verify the excellent classification accuracy of this system even for noisy or corrupted data.