Segmentation based image compression system

• US 5,586,200 A
• Filed: 01/07/1994
• Issued: 12/17/1996
• Est. Priority Date: 01/07/1994
• Status: Expired due to Term

• Current Assignee: Panasonic Corporation Of North America (Panasonic Holdings Corporation)
• Original Assignee: Panasonic Technologies, Inc. (Panasonic Holdings Corporation)
• Inventors: Westerink, Peter, Devaney, Patrick, Topper, Robert, Gnanaprakasam, Daniel
• Primary Examiner(s): Mancuso, Joseph
• Assistant Examiner(s): DEL ROSSO, GERARD DOMNICK
• Application Number:
  US08/178,948
• Time in Patent Office: 1,075 Days
• Field of Search: 382/232, 382/233, 382/234, 382/235, 382/239, 382/240, 382/244, 348/437, 348/438, 348/423, 348/397, 348/400
• US Class Current: 382/232
• CPC Class Codes: G06T 7/13 Edge detection
• CPC Class Codes: G06T 9/20 Contour coding, e.g. using detection of edges

Abstract

An image processing system encodes a natural image into a segmented or mosaic image having well-defined edges and a residual image. The segmented image is encoded using a lossless encoding technique while the residual image is encoded using a lossy technique. This encoded image may be recorded on a video tape such that the segmented image may be recovered in picture-in-shuttle modes such as fast forward and fast rewind. In addition, the recorded image may be decoded and reencoded through several generational levels without experiencing significant degradation in perceived image quality. The segmented image is produced by an encoder which employs a multi-scale edge finder that is able to resolve transitions occurring over two, four and eight pixels into an edge located between two pixels. In addition, the lossless encoder includes circuitry which chain-encodes the segmented image into a low-resolution image and multiple complementary images such that the low-resolution image may be stored as a single data packet and, thus, recovered as a unit from the tape in picture-in-shuttle mode.

Claims

1. Apparatus for compressing digital data representing an input image, the data representing the input image as a matrix of picture elements (pixels), the apparatus comprising:

processing means for processing the data representing the input image to generate data representing a segmented image having boundaries which correspond to boundaries in the input image;

means for combining the data representing the segmented image with the data representing the input image to produce data representing a residual image, the residual image representing the input image minus the segmented image using a first quantity of data;

first encoding means for losslessly encoding the data representing the segmented image to produce encoded segmented image data in a manner that allows the encoded segmented image data to be decoded without losing detail in the decoded segmented image;

second encoding means for encoding the data representing the residual image to produce encoded residual image data which represents the residual image using a second quantity of data where the first quantity of data is greater than the second quantity of data; and

means for combining the encoded segmented image data and the encoded residual image data to produce the compressed digital data.

2. Apparatus according to claim 1, wherein:

the processing means comprises means for graph-based processing the input image to produce the segmented image; and

the first encoding means comprises means for chain encoding the segmented image to produce the encoded segmented image.

3. Apparatus according to claim 1 wherein the first encoding means comprises:

means for processing the data representing the segmented image to produce a description of the segmented image as a sequence of encoded segment borders, each border being represented by a sequence of displacements that define a path around the border, to produce data representing a partially encoded segmented image;

means for encoding the data representing the partially encoded segmented image to produce a hierarchically encoded segmented image, in which each pair of values in each sequence of encoded borders of the processed segmented image is represented by a single sequence value and a complementary code value in the hierarchically encoded segmented image; and

means for processing the complementary code values and the single sequence values using a predetermined set of variable length codes to produce variable-length encoded image data as the encoded segmented image data.

4. Apparatus according to claim 3, wherein the processing means comprises:

pixel-based processing means for processing the data representing the input image to produce pixel-based image data representing the input image as a plurality of groups of pixels separated by a plurality of edges;

means for converting the data produced by the pixel-based processing means into graph data having nodes and links connecting the nodes to form connected regions, where each of the nodes represents a region in the pixel-based image data and each link represents a border between two corresponding groups of pixels in the pixel-based image data;

graph-based processing means for combining selected ones of the connected regions using the graph data to produce data representing the input image as a reduced number of connected regions; and

means for converting the data produced by the graph-based processing means into a matrix of pixel values to produce the data representing the segmented image.

5. Apparatus according to claim 4, wherein the pixel based processing means includes:

edge finder means for processing the data representing the input image to produce data representing edges in the input image; and

pixel combining means for processing the data produced by the edge finder means to generate data representing the input image as the plurality of groups of pixels separated by the plurality of edges.

6. Apparatus according to claim 5, wherein the pixel combining means includes means for combining selected groups of pixel values as defined by the data produced by the edge finder means to reduce the number of groups of pixels used to represent the input image.

7. Apparatus according to claim 5, wherein the edge finder means includes:

means for determining a first set of difference values representing respective differences in magnitude between each pixel in the image and each adjacent pixel in the matrix of pixel values;

means for determining a second set of difference values representing respective differences in magnitude between each pixel in the image and N-1 pixels in a respective group of N pixels, where each group of N pixels defines a vector on the matrix of pixels; and

means for comparing each difference value in the first set of difference values with a respective difference value in the second set of difference values to determine the greater thereof and for providing the determined values as a third set of difference values, which third set of difference values defines the edges in the input image.

8. Apparatus for compressing digital data representing an input image, the data representing the input image as a matrix of picture elements (pixels), the apparatus comprising:

processing means for processing the data representing the input image to generate data representing the image as a segmented image having groups of pixel values separated by edges where the edges correspond to boundaries in the input image;

means for subtracting the data representing the segmented image from the data representing the input image to produce data representing a residual image;

lossless encoding means for encoding the data representing the segmented image to produce encoded segmented image data in a manner that allows the encoded segmented image data to be decoded without losing any detail in the decoded segmented image;

lossy encoding means for encoding the data representing the residual image to produce encoded residual image data which represents an approximation of the residual image; and

means for combining the encoded segmented image data and the encoded residual image data to produce the compressed digital data.

9. Apparatus according to claim 8, wherein the lossless encoding means includes means for encoding the data representing the segmented image as a first component having a predetermined maximum size and as a second component representing a difference between the data representing the segmented image and the decoded first component.

10. A method of compressing digital data representing an input image, the data representing the input image as a matrix of picture elements (pixels), the method comprising the steps of:

a) locating boundaries in the input image to generate data representing a segmented image having boundaries which correspond to boundaries in the input image;

b) subtracting the data representing the segmented image from the data representing the input image to produce data representing a residual image;

c) losslessly encoding the data representing the segmented image to produce encoded segmented image data which may be decoded without losing detail in the decoded segmented image;

d) encoding the data representing the residual image to produce encoded residual image data which represents an approximation of the residual image; and

e) combining the encoded segmented image data and the encoded residual image data to produce the compressed digital data.

11. A method according to claim 10, further comprising the step of processing the data representing the segmented image, provided by step a), to combine image segments and to generate output data representing the input image as a smaller number of segments, said output data being processed by step b).

12. The method according to claim 10, wherein:

step a) further comprises the step of graph-based processing the input image to produce the segmented image; and

step c) further comprises the step of chain encoding the segmented image to produce the encoded segmented image.