An Optimal Quantizer for Bayer Pattern Image Codec ⋆

A Bayer pattern image codec method based on Wyner-Ziv structure is proposed in this paper. At the encoder the pixel of Bayer pattern image is converted into four component images by means of structural separation and transformation. Three of them are transformed into frequency domain by the discrete cosine transform. We extended the Lloyd iterative algorithm and designed an optimal scalar quantizer for the distributed codec of Bayer pattern image based on the convergence of Lagrange cost function. The luminance component is used as side information effectively, and the joint decoding and reconstruction of Baer template image is performed at the decoder. Experimental results show that this method can reduce the encoding complexity at high rates and furthermore the rate distortion performance is better improved.