Method and apparatus for coding an image object of arbitrary shape

Abstract

It is known to represent a square block of an image which has a maximum horizontal or vertical extension N as N vectors each representing one column or row, and encode it using an N×N discrete cosine transform matrix. We introduce a technique which transforms an input vector, or array, representing a portion of an arbitrarily-shaped object block in which the maximum horizontal or vertical extension is at most N, by adding to it additional padding elements, to produce a vector or array which can be encoded using an N×N discrete cosine transform. The padding maximises the number of transformed elements which are zero. We propose encoding arbitrarily-shaped objects based on this padding technique.

Description

[0001]Notice: More than one reissue application has been filed for the reissue of U.S. Pat. No. 6,718,066. The present application is a Reissue Continuation Patent Application of U.S. Reissue patent application Ser. No. 11 / 400,072, filed Apr. 6, 2006, entitled “METHOD AND APPARATUS FOR CODING AN IMAGE OBJECT OF ARBITRARY SHAPE,” now abandoned, which is a re-issue application of U.S. Pat. No. 6,718,066 (patent application Ser. No. 09 / 637,883), filed Aug. 14, 2000 and assigned to the assignee of the presently claimed subject matter.FIELD OF THE INVENTION[0002]The present invention relates to methods and apparatus for padding one or more input vectors, or a data array, representing an image object by adding components to it, to generate vector(s) or an array which can be encoded using a transform, such as a discrete cosine transform (DCT). The invention further relates to methods and apparatus for encoding the image object using the transformed vector(s) or array.BACKGROUND OF THE INVE...

AI Technical Summary

Benefits of technology

[0023]The present invention aims to provide a transform of an arbitrarily-shaped object using a normal N×N DCT array, where N is a power of 2. This makes it possible to implement the invention using presently available DCT chipsets, instead of having to design specific ICs as in the standard SA-DCT algorithm.

[0029]Optionally, the positions of the padding elements within the padded vector may be determined according to the value of K, so as to improve the efficiency of the subsequent coding.

[0033]As pointed out above, since standard SA-DCT employs not just an 8×8 DCT matrix, but also DCT matrices for each value of K, it cannot be implemented using known chipsets which implement 8×8 DCT in a highly efficient manner. By contrast, the present invention can be performed using a known chipset since it uses only an 8×8 DCT. It is true that to produce the padding elements a calculation is required, and this would usually require a new chipset, but the degree of computation required to compute the padding elements may be much less (in terms of both the number of multiplications and of additions), and therefore is easier to design. By constrast designing DCT hardware is very complicated, especially when fast algorithms are to be used.

[0034]A further advantage of the present method is that it has a nice asymmetric property between the encoder and the decoder, in that the calculation of the padding vector occurs only in the encoder, and thus no new hardware is required in the decoder. In most applications, the decoder usually has a much larger (consumer) market.

Problems solved by technology

In many such applications the available bandwidth is limited, so some form of compression is required.

A disadvantage of standard SA-DCT is that since standard SA-DCT employs not just an 8×8 DCT matrix, but also DCT matrices for each value of K, it cannot be implemented using known chipsets which implement 8×8 DCT in a highly efficient manner.

This is undesirable: the data set has been expanded after DCT, leading to large reconstruction errors if big quantization steps are used to force some of the high frequency components to be zero, or high bit rate if fine quantization steps are used to keep the reconstruction error small.

Images