A Multi-plane Fuzzy-possibilistic Net in Wavelet Transform with Vector Quantization to Color Image Compression

J.-S. Lin, S.-H. Liu, and Y.-T. Huang (Taiwan)


Color Image Compression, Hopfield neural network; Wavelet transform; Vector Quantization; Fuzzy-Possibilistic C-means.


In this paper a multi-plane unsupervised neural network called Fuzzy-Possibilistic Hopfield Net (FPHN) is created and applied to color image compression based on Vector Quantization (VQ) in Discrete Wavelet Transform (DWT). In the FPNN the fuzzy-possibilistic c-means (FPCM) is embedded into a 3-plane Hopfield neural network. First a color image is decomposed into 3-D pyramid structure with various frequency bands transformed by DWT. Then the FPHN is used to create different codebooks for these bands. The energy function of FPHN is defined as the fuzzy membership grades and possibilistic typicality degrees between training vectors and codevectors based on VQ. Finally, near global-minimum codebooks in frequency domain can be obtained when the energy function converges to a stable state. Instead of one state in a neuron for the conventional Hopfield nets, each neuron occupies 2 states called membership state and typicality state in the proposed 3-plane FPHN. The simulated results show that more valid and promising codebooks can be generated by using the proposed 3-plane FPHN than those produced by the Linde-Buzo-Gray (LBG) technique and Fuzzy Hopfield Neural Network (FHNN) in the DWT domain.

Important Links:

Go Back