Image coding method and apparatus

Claim 2

The method of claim 1, wherein said step of coding employs a variable-length code in which a non-zero value and any consecutively preceding zero values are encoded by a single codeword having a length that decreases as the consecutively preceding zero values become fewer.

Claim 3

The method of claim 1, wherein said wavelet transform is separable.

Claim 4

The method of claim 1, wherein said wavelet transform is inseparable.

Claim 5

The method of claim 1, wherein said image is rectangular.

Claim 6

The method of claim 1, wherein: said image is not rectangular, but is treated as being embedded in a rectangular area which is partitioned into image blocks as described in claim 1, and is accompanied by shape information indicating which parts of each image block belong to said image; in said step of scanning, elements which are missing from a wavelet block because said image does not occupy all of the image block from which said wavelet block is obtained are omitted from said linear sequence; and in determining said linear sequence, said elements which are missing are treated as having values of zero.

Claim 7

The method of claim 1, wherein each said wavelet block has just one purely low-frequency element.

Claim 8

The method of claim 1, wherein all purely low-frequency elements and the horizontal, vertical, and diagonal high-frequency elements having lowest spatial frequencies constitute a low-frequency group, and among the high-frequency elements having higher spatial frequencies, the horizontal elements constitute a horizontal group, the vertical elements constitute a vertical group, and the diagonal elements constitute a diagonal group, said step of scanning comprising the further steps of: scanning at least part of said horizontal group, if said low-frequency group has a non-zero horizontal element, and scanning at least part of said vertical group, if said low-frequency group has a non-zero vertical element; then scanning at least part of said diagonal group, if said low-frequency group has a non-zero diagonal element that is not a purely low-frequency element; then scanning at least part of said horizontal group, if said low-frequency group has a zero-valued horizontal element, and scanning at least part of said vertical group, if said low-frequency group has a zero-valued vertical element; then scanning at least part of said diagonal group, if said low-frequency group has a zero-valued diagonal element that is not a purely low-frequency element.

Claim 9

The method of claim 8 wherein: when a non-zero horizontal element is scanned in said low-frequency group, at least part of said horizontal group is scanned as recited in claim 8 before further elements in said low-frequency group are scanned; and when a non-zero vertical element is scanned in said low-frequency group, at least part of said verticaL group is scanned as recited in claim 8 before further elements in said low-frequency group are scanned.

Claim 10

The method of claim 8, wherein said horizontal group, said vertical group, and said diagonal group are divided into two-by-two subblocks, and in said step of scanning, when any one of said two-by-two subblocks is scanned, all elements in that two-by-two subblock are scanned before any elements in any other two-by-two subblock scanned.

Claim 11

The method of claim 8, wherein: within each group among said horizontal group, said vertical group, and said diagonal group, when a non-zero element is scanned, a related element with a higher spatial frequency is scanned next if possible, and when a zero-valued element is scanned, a related element with a lower spatial frequency is scanned next if possible.

Claim 12

The method of claim 8, wherein elements that have a highest spatial frequency in said diagonal group and are related to zero-valued lower-frequency elements in said diagonal group are scanned last among all elements in said wavelet block.

Claim 13

The method of claim 8 wherein, among elements with identical spatial frequencies, elements related to lower-frequency non-zero elements are scanned before elements related to zero-valued lower-frequency elements.

Claim 14

An image compressor for compressing a digitized image, comprising: a block partitioner (2) for partitioning said digitized image into image blocks; a wavelet transform processor (4) coupled to said block partitioner (2), for performing a wavelet transform on each image block, thereby obtaining, for each image block, a wavelet block having at least one purely low-frequency element, a plurality of diagonal high-frequency elements representing equal horizontal and vertical spatial frequencies, a plurality of horizontal high-frequency elements representing higher horizontal than vertical spatial frequencies, and a plurality of vertical high-frequency elements representing higher vertical than horizontal spatial frequencies; a quantizer (6) coupled to said separable wavelet transform processor (4), for quantizing the elements in each said wavelet block; an adaptive scanner (8) coupled to said quantizer (6), for scanning each said wavelet block as quantized by said quantizer (6), starting with said purely low-frequency element, and scanning the horizontal, vertical, and diagonal high-frequency elements in a sequence responsive to occurrence of non-zero elements among said horizontal, vertical, and diagonal high-frequency elements, thereby rearranging the elements in each said wavelet block in a linear sequence; and an encoder (10) coupled to said adaptive scanner (8), for coding the elements in each said wavelet block, as quantized by said quantizer (6), according to said linear sequence.

Claim 15

The apparatus of claim 14, wherein said encoder (10) employs a variable-length code in which a non-zero value and any consecutively preceding zero values are encoded by a single codeword having a length that decreases as the consecutively preceding zero values become fewer.

Claim 16

The apparatus of claim 14, wherein said wavelet transform is separable.

Claim 17

The apparatus of claim 14, wherein said wavelet transform is inseparable.

Claim 18

The apparatus of claim 14, wherein said image is rectangular.

Claim 19

The apparatus of claim 14, wherein: said image is not rectangular, but is treated as being embedded in a rectangular area which is partitioned into image blocks as described in claim 14, and is accompanied by shape information indicating which parts of each image block belong to said image; said adaptive scanner (8) skips elements which are missing from a wavelet block because said image does not occupy all of the image block from which said wavelet block is obtained; and in determining said linear sequence, said adaptive scanner (8) treats said elements which are missing as having values of zero.

Claim 20

The apparatus of claim 14, wherein each said wavelet block has just one purely low-frequency element.

Claim 21

The apparatus of claim 14, wherein all purely low-frequency elements and the horizontal, vertical, and diagonal high-frequency elements having lowest spatial frequencies constitute a low-frequency group, and among the high-frequency elements having higher spatial frequencies, the horizontal elements constitute a horizontal group, the vertical elements constitute a vertical group, and the diagonal elements constitute a diagonal group, and wherein, in scanning each said wavelet block, said adaptive scanner (8): scans at least part of said horizontal group, if said low-frequency group has a non-zero horizontal element, and scans at least part of said vertical group, if said low-frequency group has a non-zero vertical element; then scans at least part of said diagonal group, if said low-frequency group has a non-zero diagonal element that is not a purely low-frequency element; then scans at least part of said horizontal group, if said low-frequency group has a zero-valued horizontal element, and scans at least part of said vertical group, if said low-frequency group has a zero-valued vertical element; then scans at least part of said diagonal group, if said low-frequency group has a zero-valued diagonal element that is not a purely low-frequency element.

Claim 22

The apparatus of claim 21 wherein: upon scanning a non-zero horizontal element in said low-frequency group, said adaptive scanner (8) proceeds to scan at least part of said horizontal group, as recited in claim 21, before scanning further elements in said low-frequency group; and upon scanning a non-zero vertical element in said low-frequency group, said adaptive scanner (8) proceeds to scan at least part of said verticaL group, as recited in claim 21, before scanning further elements in said low-frequency group.

Claim 23

The apparatus of claim 21, wherein said horizontal group, said vertical group, and said diagonal group are divided into two-by-two subblocks, and said adaptive scanner (8) scans each one of said two-by-two subblocks completely before proceeding to scan any other one of said two-by-two subblocks.

Claim 24

The apparatus of claim 21, wherein: within each group among said horizontal group, said vertical group, and said diagonal group, upon scanning a non-zero element, said adaptive scanner (8) proceeds if possible to scan a related element with a higher spatial frequency, and upon scanning a zero-valued element, said adaptive scanner (8) proceeds if possible to scan a related element with a lower spatial frequency.

Claim 25

The apparatus of claim 21, wherein said adaptive scanner 8 scans last, among all elements in said wavelet block, the diagonal elements that have a highest spatial frequency in said diagonal group and are related to zero-valued lower-frequency elements in said diagonal group.

Claim 26

The apparatus of claim 21 wherein, among elements with identical spatial frequencies, said adaptive scanner (8) scans elements related to lower-frequency non-zero elements before scanning elements related to zero-valued lower-frequency elements.