| 摘要 |
Hierarchal encoding of an image is achieved by first deriving a picture element array of binary coded intensity values representing the image and then deriving from the intensity values of the picture element array a sequence of hierarchal codes consisting of one of the intensity values as the first code followed by n-1 cyclic differences between n-1 distinct pairs of intensity values, where n is the total number of intensity values of the picture element array. The distinct pairs of intensity values are systematically selected by making alternating horizontal and vertical passes through the picture element array of intensity values. In each horizontal pass, consecutive nonoverlapping adjoining pairs of intensity values are selected along consecutive rows of the picture element array and successive undersampled arrays, the undersampled array for a horizontal pass being constructed by eliminating every other row from the undersampled array for the preceding vertical pass. In each vertical pass consecutive vertically adjoining pairs of intensity values are selected along consecutive nonoverlapping adjacent pairs of rows of successive undersampled arrays, the undersampled array of a vertical pass being constructed by eliminating every other column from the picture element array or the undersampled array for the preceding horizontal pass. The sequence of hierarchal codes may be compressed with variable length coding prior to transmission or storage. Decoding of the sequence of hierarchal codes to reconstruct the picture element array of intensity values follows the inverse of the encoding procedure, in which the first code of the sequence is the intensity value of a designated position of the picture element array, and the other intensity values are derived by taking the cyclic difference between decoded intensity values and successive hierarchal codes in alternating horizontal and vertical passes corresponding to those used in the encoding procedure.
|
