Browse Prior Art Database

OPTIMIZED DECODING FOR IMAGE COMPRESSION ALGORITHMS

IP.com Disclosure Number: IPCOM000009449D
Original Publication Date: 1999-Sep-01
Included in the Prior Art Database: 2002-Aug-26
Document File: 4 page(s) / 134K

Publishing Venue

Motorola

Related People

Manish Singhal: AUTHOR [+2]

Abstract

Most of Image Compression standards and algo- rithms (for photographic as well as moving pic- tures), e.g. H.261, H.263, JPEG, MPEGl,2,4 etc. use Discrete Cosine Transforms (DCT) to remove spatial redundancy. These algorithms use DCT and Inverse Discrete Cosine Transforms (IDCT) mod- ules in an encoder, and a IDCT module in a decoder. The IDCT module is one of the computation inten- sive blocks in the encoder as well as in the decoder. This paper exploits the low frequency content of the input to an IDCT module to increase the computa- tion speed of the encoder and the decoder. Various simulations have indicated significant improvement in the computation speed of the encoder and the decoder.

This text was extracted from a PDF file.
At least one non-text object (such as an image or picture) has been suppressed.
This is the abbreviated version, containing approximately 50% of the total text.

Page 1 of 4

Technical MOTOROLA @ Developments

   OPTIMIZED DECODING FOR IMAGE COMPRESSION ALGORITHMS

by Manish Singhal and Bhaskar Bhupalam

simulations have indicated significant improvement in the computation speed of the encoder and the decoder.

BACKGROUND

  Description of a generic decoding loop. First, the input image or prediction error image is divided into 8x8 blocks. Then by taking the DCT of the block, the block gets transformed into the frequency domain. The block is quantized and coded. In the decoder, and the encoder feedback loop, IDCT of the block is taken to get the block data back to the time domain. This is represented in Figures 1 and 2.

w For DCT further Encoding

Fig. 1 Decoding Loop in Encoder

0 ?dammla, 1°C. ,999 September 1999

ABSTRACT

  Most of Image Compression standards and algo- rithms (for photographic as well as moving pic- tures), e.g. H.261, H.263, JPEG, MPEGl,2,4 etc. use Discrete Cosine Transforms (DCT) to remove spatial redundancy. These algorithms use DCT and Inverse Discrete Cosine Transforms (IDCT) mod- ules in an encoder, and a IDCT module in a decoder. The IDCT module is one of the computation inten- sive blocks in the encoder as well as in the decoder. This paper exploits the low frequency content of the input to an IDCT module to increase the computa- tion speed of the encoder and the decoder. Various

[This page contains 14 pictures or other non-text objects]

Page 2 of 4

Developments Technical 0 M MOTOROLA

r - - - - - - - - -~~- - ~-"- - - 7

              I I After preliminaq 1

decoding I

' For further

) IDCT I I decoding I &Display

L---------------J

Fig. 2 Decoding Loop in Decoder

  The detailed flow graph of the Decoding Loop is INTRA coding. In INTRA coding the inverse trans- shown in Figure 3. The Quantized DCT coefficients formed data is clipped normally to [0:255]. In are Inverse Quantized and Inverse Discrete Cosine INTER coding, the decoded prediction error is transformed. If the input to the DCT is prediction added with its prediction to reconstruct the image error (with reference to some other image), it is and is then clipped.
called an INTER coding otherwise it is called an

1 Quantized DCT Data

No

Reconstruct

F Clip

Reconstructed Image

Fig. 3 Flow Chart for Decoding Loop

0 Motomla, Inc. ,999 87 September 1999

[This page contains 14 pictures or other non-text objects]

Page 3 of 4

Developments Technical 0 M MOlOROLA...