A noval novel technique is described for high speed color conversions for JPEG-compressed images. The color conversion is processed in the transform domain instead of the traditional spatial domain. Only the input DC coefficients of the multiple color components are processed through the traditional table lookup to create output DC coefficients. Given each output color component's DC value, the linearity of its 63 AC coefficients is determined and their transform domain scaling functions are looked up in a 1-D table as a function of only the component output DC term. For n-dimensional input color space to m-dimensional output color space conversion, n component blocks create m output blocks. These blocks can be inverse DCTed to create spatial domain data or requantized and entropy encoded to create JPEG compressed data. Abort criteria also have been developed to determine when the traditional spatial domain interpolation needs to be done.
展开▼
