摘要:
In accordance with the present invention, the R,G,B color image signals from a single sensor having a color filter array are all transformed to .GAMMA.-space by changing them to R.sup.1/.GAMMA., G.sup.1/.GAMMA., B.sup.1/.GAMMA., respectively, where .GAMMA. is approximately 2.4. In this space, all operations such as color differencing, interpolation of those missing pixels required for color differencing, compression, decompression, edge enhancement and final interpolation of all missing pixels are performed without further transformation of the image signals. For the same final bit rate, noise in the reproduced image is reduced by refraining from interpolating the missing color pixels prior to compression of the image data. In order to avoid over-emphasizing features of the image which are already sufficiently sharp, the combined outputs of horizontal and vertical sharpening processes are subjected to a paring process of the invention which suppresses strong high-spatial frequency components as a function of their amplitude. In the compression-decompression process of the invention, each spatial frequency coefficient of the spatial frequency-transformed image is divided by a normalization factor determined by cascading in the spatial frequency domain the human visual system contrast sensitivity function, the edge enhancement modulation transfer function and the image display modulation transfer function and inversing the resulting matrix elements.