AU751069B2

AU751069B2 - Method and device for storing at least one digital image

Info

Publication number: AU751069B2
Application number: AU97341/98A
Authority: AU
Inventors: Christoph Heer
Original assignee: Siemens AG; Siemens Corp
Current assignee: Siemens AG
Priority date: 1997-09-29
Filing date: 1998-08-18
Publication date: 2002-08-08
Anticipated expiration: 2018-08-18
Also published as: JP2001518763A; EP1020082A1; AU9734198A; WO1999017551A1; KR20010015667A

Description

GR 97 P 2675 Foreign version Description Method and apparatus for storage of at least one digital image The invention relates to the storage of digital images.

Such a method is known for a method for blockbased image coding from In this method, the digital images which are recorded by a camera have at least one image object. The image objects are of any shape, which is indicated in the form of contour information associated with the image object.

The pixels are furthermore grouped into image blocks, normally with a size of 8x8 or 16x16 pixels.

In the method known from the following procedure is described for storage of the image objects in the image. Each image object is described by a rectangle which surrounds the image object and is broken down into rectangular blocks (8x8 or 16x16 pixels). The image blocks within the circumscribing rectangle are investigated to determine whether the image blocks in the rectangle all contain pixels which are or are not associated with the image object. Image blocks which are not associated with the image objectare marked and are written to the memory with a value which can be predetermined, for example a coding value with the value 255 for a black pixel, or with the value 0 for a white pixel. This procedure leads to a considerable memory requirement for the storage of image objects.

In the method known from for object-based image coding, the contour information which is in each case used to describe an image object is used only for masking the rectangular blocks for the object-based processing steps.

GR 97 P 2675 Foreign version 2 The contour information exists, for example, in socalled alpha planes.

An alpha plane is a matrix with binary values, with a first binary value for the respective pixel in the image indicating that the pixel is associated with the image object, and a second binary value for the respective pixel in the image indicating that the pixel is not associated with the image object.

M4R coding for the intracoded image objects is also known from in which the contour information for the respective image object is likewise included.

A method for image coding, in which all the edges are extracted from an image, is known from The extracted edges are stored in an edge file. The extracted edges are furthermore subjected to processing, and the processed edges are removed from the image, thus resulting in a difference image. The difference image as well as the edge data are coded and transmitted.

An object-oriented image coding method is known from The invention is based on the problem of specifying a method as well as an apparatus by means of which it is possible to store image objects with a memory space requirement which is less than that for the known methods.

The problem is solved by the method as claimed in patent claim 1, and by the apparatus as claimed in patent claim 6.

In the method, at least some of the pixels are grouped into at least one image object. The pixels are furthermore grouped into image blocks. Contour information of the image object is associated with the image. The image is stored GR 97 P 2675 Foreign version 3 taking account of the contour information of the image object.

The apparatus has a processor unit which is set up in such a manner that the image is stored taking account of the contour information. The contour information is associated with the image. At least some of the pixels are grouped into the image object, and the pixels are furthermore grouped into image blocks.

The invention results in a considerable saving in memory space requirement, and this is a considerable advantage, particularly in the case of object-based image coding methods.

Advantageous developments of the invention result from the dependent claims.

In one development of the invention, it is advantageous to provide a memory header field for an image object, in which information is contained by means of which the stored pixels can be uniquely associated with their physical position in the image.

In this case, the only pixels stored are those which are associated with the image object.

This development describes a very efficient method for storage of the information for an image object. The figures show an exemplary embodiment of the invention, which will be explained in more detail in the following text, and in which: Figure 1 shows a sketch illustrating the exemplary embodiment graphically; Figures 2a and 2b show a sketch of an image coder unit as well as an image decoder unit using the MPEG4 method.

Figures 3a and 3b show sketches of an apparatus for storage (Figure 3a) of, and for reading (Figure 3b), an image block GR 97 P 2675 Foreign version -4taking account of the contour information; A sequence of images 201 recorded by a camera is coded in an image coder unit 200 (see Figure 2a) to form an image data stream 202.

In this case, the image which has pixels is grouped into image blocks with a size of 8x8 or 16x16 pixels. Furthermore, one image has at least one image object. The image object 101 (see Figure 1) is of any desired shape and has any desired number of pixels 102 associated with the image object 101.

If the image 100 contains a plurality of image objects 101, then the image objects are separately coded using the method according to the H.263 method for block-based hybrid DCT image coding. This is done in means 203, 204, 205 for coding the image objects, to which the individual image objects 210, 211, 212, respectively, are passed.

The coded image objects 206, 207, 208 are passed to a multiplexer 209, which multiplexes the coded image objects to form the image data stream 202.

The image data stream 202 is transmitted using any desired transmission system from the image coder unit 200 to an image decoder unit 250. Once the image data stream 202 has been received in the image decoder unit 250, the image data stream 202 is split, in a demultiplexer 251, into the individual coded image objects 252, 253, 254. The image objects are respectively decoded in a means for decoding 255, 256, 257 an image object, and the decoded image objects 258, 259, 260 are reconstructed to form the image 100, in a means 261 for reconstruction of the images (see Figure 2b).

GR 97 P 2675 Foreign version 5 Coding information is assigned to the pixels, that is to say luminance information and chrominance information is assigned to the pixels in the images 201, and the contour information for the respective image object, that is to say parameters which indicate the shape of the image object, is/are assigned to each image object 101.

The contour information exists in the form of a matrix which has binary values. A first binary value in the matrix indicates that the respective pixel which represents the value in the matrix is associated with the image object. A second binary value indicates that the respective pixel is not associated with the image object.

This contour information is associated with the respective image object during the coding process.

In the past, both the image blocks which are associated with the image object 101 as well as the image blocks 103 which are not associated with the image object 101 were stored, the image objects which are not associated 103 with the image object [lacuna] with a default value 104.

According to the invention, the image object 101 is stored in a memory 262, taking account of the contour information.

In order to store 8x8 image blocks, a memory header field is provided in which 1 bit in the memory header field 105, which indicates the association with the image object, is provided in the form of a matrix for each pixel in the area of the image blocks. The only coding information values from the pixels which are stored in the memory 202 are those which are actually associated with the image object 101.

The storage process takes place in a sequence which can be predetermined, line-by-line, column-bycolumn, or using a zigzag scan. The matrix (8x8 bit matrix) with a size of 8 bytes is interrogated line-byline, GR 97 P 2675 Foreign version 6 so that the value of the coding information associated with the pixel can be determined uniquely for each pixel.

In the matrix, a first binary value is once again assigned for a pixel which is associated with the image object 101, and the second binary value is assigned to a pixel which is not associated with the image object 101. Furthermore, the length of the stored image data for the respective image object is indicated in the memory header field 105.

Figure 3a shows a sketch of an apparatus for storage of an image block, taking account of the contour information.

The contour information in the image block 301 to be stored is stored as a bit matrix in a field 302 which can be written to in parallel (8 bits per line) and can be read in serial form. A processor unit 303 controls the process of storing the image block 301 and its contour information via a multiplexer 304, for selection of pixel data relating to the image block 301 or the bit matrix from the field 302, and a unit for address calculation 305 for the memory area 306.

The data are stored in serial form. First of all, the 8 lines of the bit matrix (8 bytes) are written from the field 302 to the addresses 3 to 10 in the memory area 306. The image block 301 is then written to the memory. Depending on the binary value of the bit matrix which is read in serial form from the field 301, the corresponding pixel is written from the image block 301 to the memory area 306, or is masked out via the multiplexer 304. If the binary value of the bit matrix has the value which indicates that the respective pixel, to which [lacuna] image object is associated, [lacuna] an address is calculated in an address calculation unit 305, and the respective pixel is stored at the appropriate address in the memory area 306. If the binary value of the bit GR 97 P 2675 Foreign version 7 matrix has the value which indicates that the respective pixel, to which [lacuna] image object is not assigned, [lacuna] no address is calculated in the address calculation unit 305, and the respective pixel in the image block 301 is jumped over. Once all the pixels in the image block 301 which are associated with the image object have been written to the memory area 306, the number of these pixels, together with an information word which indicates the binary coding, is written by the processor unit, via the multiplexer 304, as header field information to the first two addresses 1 and 2 in the memory area 306. The data corresponding to the object data, as described in Figure 1, are present in the memory area.

Figure 3b shows a sketch of an apparatus for reading an image block which has been stored taking account of the contour information.

Header field information relating to the image block stored in the memory area 350 is read, and is decoded in a processor unit 351. The processor unit uses the address calculation unit 352 to control the reading of the next 8 data words, which are stored, via a multiplexer 353, as a bit matrix in a field 354 which can be written to in parallel (8 bits per line) and can be read in serial form. The number of data words which are stored in the memory area 350, that is to say those which are associated with the image block to be read, is decoded from the header field information, in the processor unit 351. These addresses are addressed via the address calculation unit 352, and the data words can be written to a queuing unit 355.

The data are read in serial form. The bit matrix is read from the field 354 in serial form.

Depending on the binary value of the bit matrix, the appropriate pixel is output from the queuing unit 355, or a standard value is output, via the multiplexer 356.

If the binary value of the bit matrix has the value GR 97 P 2675 Foreign version 7a which indicates that the respective pixel, with which [lacuna] GR 97 P 2675 Foreign version -8 image object is associated, [lacuna] a data item is output from the queuing unit 355. If the binary value of the bit matrix has the value which indicates that the respective pixel, to which [lacuna] image object is not assigned, [lacuna] a standard value which can be predetermined is output. If the standard value is output, the present value is maintained at the output of the queuing unit 355.

GR 97 P 2675 Foreign version 9 The following publications have been cited in the course of this document: ISO/IEC JTC1/SC29/WG11, Coding of moving pictures and associated audio information, MPEG4 Video Verification Model Version 1.0, January 1996 US 5 615 287 B1 EP 0 825 556 Al

Claims

1. A method for storing a digital image which has pixels, the method comprising the steps of: grouping some of the pixels into an image object; grouping pixels into image blocks; associating contour information of the image object with some of the pixels; determining a reduced image data set using the contour information; and storing the image using the reduced image data set. o o

2. The method as claimed in claim 1, in which the contour information is provided in the form of a matrix with binary values, with a first binary value for the respective pixel in the image indicating that the pixel is associated with the image object, and a second binary value for the respective pixel in the image indicating that the pixel is not associated with the image object.

3. The method as claimed in claim 1 or 2, in which the image has a plurality of image objects. S4. The method as claimed in one of claims 1 to 3, in which the only pixels stored are those which are associated with the image object.

5. The method as claimed in claim 4, in which a memory header field is provided for an image object, in which information is contained by means of which the stored pixels can be uniquely associated with their physical position in the image. \LIBE]03747.doc:cdg -11-

6. An apparatus for storing a digital image which has pixels, the apparatus having a processor unit and comprising: means for grouping some of the pixels into an image object; means for grouping pixels into image blocks; means for associating contour information of the image object with some of the pixels; means for determining a reduced image data set using the contour information; and means for storing the image using the reduced image data set. eeeee

7. The apparatus as claimed in claim 6, in which the processor unit is set up in such a manner that the contour information is provided in the form of a matrix with binary values, with a first binary value for the respective pixel in the image indicating that the pixel is associated with the image object, and a second binary value for the respective pixel in the image indicating that the pixel is not associated with the image object.

8. The apparatus as claimed in claims 6 or 7, in which the processor unit is set up in S:..such a manner that the image has a plurality of image objects.

9. The apparatus as claimed in one of the claims 6 to 8, in which the processor unit is set up in such a manner that the only pixels stored are those which are associated with the image object. [R:\LI B E]03747.doc:edg

12- The apparatus as claimed in claim 9, in which the processor unit is set up in such a manner that a memory header field is provided for an image object, in which information is contained by means of which the stored pixels can be uniquely associated with their physical position in the image. DATED this Thirtieth Day of May, 2002 Siemens Aktiengesellschaft Patent Attorneys for the Applicant SPRUSON FERGUSON [R:\LIBE]03747.doc:edg