## Improvement of DCT-based compression algorithms using Poisson’s equation (2006)

Venue: | IEEE Trans. Image Proc |

Citations: | 9 - 5 self |

### BibTeX

@ARTICLE{Yamatani06improvementof,

author = {Katsu Yamatani and Naoki Saito and Senior Member},

title = {Improvement of DCT-based compression algorithms using Poisson’s equation},

journal = {IEEE Trans. Image Proc},

year = {2006},

volume = {15},

pages = {3672--3689}

}

### OpenURL

### Abstract

We propose two new image compression-decompression methods that reproduce images with better visual fidelity, less blocking artifacts, and better PSNR, particularly in low bit rates, than those processed by the JPEG Baseline method at the same bit rates. The additional computational cost is small, i.e., linearly proportional to the number of pixels in an input image. The first method, the “full mode ” polyharmonic local cosine transform (PHLCT), modifies the encoder and decoder parts of the JPEG Baseline method. The goal of the full mode PHLCT is to reduce the code size in the encoding part and reduce the blocking artifacts in the decoder part. The second one, the “partial mode ” PHLCT (or PPHLCT for short), modifies only the decoder part, and consequently, accepts the JPEG files, yet decompresses them with higher quality with less blocking artifacts. The key idea behind these algorithms is a decomposition of each image block into a polyharmonic component and a residual. The polyharmonic component in this paper is an approximate solution to Poisson’s equation with the Neumann boundary condition, which means that it is a smooth predictor of the original image block only using the image gradient information across the block boundary. Thus the residual—obtained by removing the polyharmonic component from the original image block—has approximately zero gradient across the block boundary, which gives rise

### Citations

927 | The JPEG Still Picture Compression Standard
- Wallace
- 1991
(Show Context)
Citation Context ...es. Although new standards such as JPEG-2000 and MPEG-21 (both of them have adopted wavelets) are emerging, many systems including digital cameras and image databases still use the JPEG standard [1], =-=[2]-=-, which is based on the Discrete Cosine Transform (DCT) [3], [4], [5]. The reasons why DCT—sometimes called DCT Type II or DCT-II [4], [5]—is used in many compression algorithms instead of the Discret... |

380 |
Discrete Cosine Transforms: Algorithms, Advantages
- Rao, Yip
- 1990
(Show Context)
Citation Context ...f them have adopted wavelets) are emerging, many systems including digital cameras and image databases still use the JPEG standard [1], [2], which is based on the Discrete Cosine Transform (DCT) [3], =-=[4]-=-, [5]. The reasons why DCT—sometimes called DCT Type II or DCT-II [4], [5]—is used in many compression algorithms instead of the Discrete Fourier Transform (DFT) or the Discrete Sine Transform (DST) a... |

323 |
Discrete cosine transform
- Ahmed, Natarajan, et al.
- 1974
(Show Context)
Citation Context ...oth of them have adopted wavelets) are emerging, many systems including digital cameras and image databases still use the JPEG standard [1], [2], which is based on the Discrete Cosine Transform (DCT) =-=[3]-=-, [4], [5]. The reasons why DCT—sometimes called DCT Type II or DCT-II [4], [5]—is used in many compression algorithms instead of the Discrete Fourier Transform (DFT) or the Discrete Sine Transform (D... |

211 | Introduction to partial differential equations (Second Edition - Folland - 1995 |

149 | Data compression and harmonic analysis
- Donoho, Vetterli, et al.
- 1998
(Show Context)
Citation Context ... signals to be compressed [6]. It is well known that KLT is the most efficient linear transform for a collection of signals if the reconstruction error is measured in the mean square error (MSE) [6], =-=[7]-=-. 2) Even more important is its treatment of the block boundaries. Taking DCT of an image block is equivalent to: a) extending the data by even reflection at the block boundaries; b) taking DFT of the... |

125 |
Remarques sur l’analyse de Fourier à fenêtre
- Coifman, Meyer
- 1991
(Show Context)
Citation Context ...cillations, Saito and Remy proposed the Polyharmonic Local Sine Transform (PHLST) [16], [17]. PHLST also compensates several problems in the local trigonometric transforms (LTTs) of Coifman and Meyer =-=[20]-=- and Malvar [21], [22], such as the overlapping windows and the slope of the bell functions. Although PHLST can be defined quite generally as in [17], we restrict ourselves to the simplest case in Feb... |

103 | Image quality assessment: From error measurement to structural similarity
- Zhou, Bovik, et al.
(Show Context)
Citation Context ...he square root of the total number of pixels. The unit of PSNR is decibel (dB). MSSIM is a more perceptually-correct measure of similarity between two images that was recently proposed by Wang et al. =-=[26]-=-. This is based on the comparison of the local patterns of pixel intensities normalized for luminance and contrast. If the original image is regarded as the one with the perfect quality, then the MSSI... |

92 |
The LOT: transform coding without blocking effects
- Malvar, Staelin
- 1989
(Show Context)
Citation Context ...o and Remy proposed the Polyharmonic Local Sine Transform (PHLST) [16], [17]. PHLST also compensates several problems in the local trigonometric transforms (LTTs) of Coifman and Meyer [20] and Malvar =-=[21]-=-, [22], such as the overlapping windows and the slope of the bell functions. Although PHLST can be defined quite generally as in [17], we restrict ourselves to the simplest case in February 17, 2006 D... |

77 | Discrete Cosine Transform
- Strang, “The
- 1999
(Show Context)
Citation Context ...m have adopted wavelets) are emerging, many systems including digital cameras and image databases still use the JPEG standard [1], [2], which is based on the Discrete Cosine Transform (DCT) [3], [4], =-=[5]-=-. The reasons why DCT—sometimes called DCT Type II or DCT-II [4], [5]—is used in many compression algorithms instead of the Discrete Fourier Transform (DFT) or the Discrete Sine Transform (DST) are th... |

40 | An optimization approach for removing blocking effects in transform coding
- Minami, Zakhor
- 1995
(Show Context)
Citation Context ...tifacts Reduction We also evaluated the performance of the reduction of blocking artifacts in the decompressed images by using a measure called the mean squared difference of slope (MSDS) proposed in =-=[27]-=- and used (with modification) in [28] and [18]. The MSDS values were computed from the sequences of four consecutive pixel values in horizontal/vertical directions across a block boundary or of four p... |

24 |
Artifact reduction in low bit rate DCT-based image compression
- Luo, Chen, et al.
- 1996
(Show Context)
Citation Context ...oticeable and annoying for human observers. Many techniques have been developed to remove the blocking artifacts while keeping the objectively measured image quality (e.g., PSNR) constant, e.g., [8], =-=[9]-=-, [10], [11], [12]. These all operate on the spatial domain, i.e., these are post-processing techniques after the decompression is done. On the other hand, another set of techniques has been suggested... |

17 | Vozovoi: \A fast Poisson solver of arbitrary order accuracy in rectangular regions
- Averbuch, Israeli, et al.
- 1998
(Show Context)
Citation Context ... [17]. The polyharmonic components can be computed quickly by utilizing the computationally-fast and numerically-accurate Laplace/Poisson solver developed by Averbuch, Braverman, Israeli, and Vozovoi =-=[23]-=-, [24] as long as the boundary data are available. Combining this feature with the quickly decaying expansion coefficients of the residuals, the usefulness of PHLST to image approximation was demonstr... |

14 | Image Information Restoration Based on Long-Range Correlation
- Wang
(Show Context)
Citation Context ...oying for human observers. Many techniques have been developed to remove the blocking artifacts while keeping the objectively measured image quality (e.g., PSNR) constant, e.g., [8], [9], [10], [11], =-=[12]-=-. These all operate on the spatial domain, i.e., these are post-processing techniques after the decompression is done. On the other hand, another set of techniques has been suggested as an option in t... |

13 |
A fast 3D Poisson solver of arbitrary order accuracy
- Braverman, Israeli, et al.
- 1998
(Show Context)
Citation Context ... The polyharmonic components can be computed quickly by utilizing the computationally-fast and numerically-accurate Laplace/Poisson solver developed by Averbuch, Braverman, Israeli, and Vozovoi [23], =-=[24]-=- as long as the boundary data are available. Combining this feature with the quickly decaying expansion coefficients of the residuals, the usefulness of PHLST to image approximation was demonstrated [... |

13 |
Derivation of prediction equations for blocking effect reduction
- Lakhani, Zhong
- 1999
(Show Context)
Citation Context ...he performance of the reduction of blocking artifacts in the decompressed images by using a measure called the mean squared difference of slope (MSDS) proposed in [27] and used (with modification) in =-=[28]-=- and [18]. The MSDS values were computed from the sequences of four consecutive pixel values in horizontal/vertical directions across a block boundary or of four pixel values in diagonal directions at... |

12 |
Relation between the Karhunen-Loève and cosine transforms
- Clarke
(Show Context)
Citation Context ...te Sine Transform (DST) are the following. 1) DCT theoretically approximates the Karhunen-Loève Transform (KLT) if we assume the first order Markov property in a given set of signals to be compressed =-=[6]-=-. It is well known that KLT is the most efficient linear transform for a collection of signals if the reconstruction error is measured in the mean square error (MSE) [6], [7]. 2) Even more important i... |

11 | Blocking artifact detection and reduction in compressed data
- Triantafyllidis, Tzovaras, et al.
- 2002
(Show Context)
Citation Context ...in the case of PPHLCT. (Note that in PHLCT we should be satisfied if the U Q predicts F as well as U does.) From the viewpoint of prediction, there are several algorithms, [1, Sec. 16.1], [13], [14], =-=[18]-=-, [15], which predict three to nine low frequency AC components of F . Therefore, in this section, we shall measure the performance of U and U Q as a predictor of F and compare the results with those ... |

3 |
A new local sine transform without overlaps: A combination of computational harmonic analysis and
- Saito, Remy
(Show Context)
Citation Context ...ge from the artificial discontinuities that cause the blocking artifacts. To address these problems, Saito and Remy recently developed a new transform called Polyharmonic Local Sine Transform (PHLST) =-=[16]-=-, [17]. PHLST decomposes the data in each block into two components: the polyharmonic component and the residual. The polyharmonic component is obtained by solving the polyharmonic equation (e.g., Lap... |

2 | Applied Analysis, Englewood Cliffs, N.J - Lanczos - 1956 |

1 |
A new approach for real time reduction of blocking effect
- Hong, Chan, et al.
- 1998
(Show Context)
Citation Context ...ite noticeable and annoying for human observers. Many techniques have been developed to remove the blocking artifacts while keeping the objectively measured image quality (e.g., PSNR) constant, e.g., =-=[8]-=-, [9], [10], [11], [12]. These all operate on the spatial domain, i.e., these are post-processing techniques after the decompression is done. On the other hand, another set of techniques has been sugg... |

1 |
Image processing method for reducing noise and blocking artifact in a digital image
- Lu, Luo, et al.
(Show Context)
Citation Context ...able and annoying for human observers. Many techniques have been developed to remove the blocking artifacts while keeping the objectively measured image quality (e.g., PSNR) constant, e.g., [8], [9], =-=[10]-=-, [11], [12]. These all operate on the spatial domain, i.e., these are post-processing techniques after the decompression is done. On the other hand, another set of techniques has been suggested as an... |

1 |
distortion corrector and image signal expander
- Kawasaka, “Block
- 1999
(Show Context)
Citation Context ...nd annoying for human observers. Many techniques have been developed to remove the blocking artifacts while keeping the objectively measured image quality (e.g., PSNR) constant, e.g., [8], [9], [10], =-=[11]-=-, [12]. These all operate on the spatial domain, i.e., these are post-processing techniques after the decompression is done. On the other hand, another set of techniques has been suggested as an optio... |

1 |
Improved image reproduction from dc components
- Lakhani
- 1996
(Show Context)
Citation Context ...re post-processing techniques after the decompression is done. On the other hand, another set of techniques has been suggested as an option in the JPEG standard [1, Sec. 16.1] and further explored in =-=[13]-=-, [14], [15]. These methods are all based on the prediction of the AC coefficients by fitting quadratic or cubic polynomial surface using the DC coefficients of several blocks. In general, polynomial ... |

1 | equations for JPEG’s blocking artifacts reduction approach - “Improved - 1997 |

1 |
Surface fitting approach for reducing blocking artifacts in low bitrate DCT decoded images
- Kieu, Nguyen
(Show Context)
Citation Context ...essing techniques after the decompression is done. On the other hand, another set of techniques has been suggested as an option in the JPEG standard [1, Sec. 16.1] and further explored in [13], [14], =-=[15]-=-. These methods are all based on the prediction of the AC coefficients by fitting quadratic or cubic polynomial surface using the DC coefficients of several blocks. In general, polynomial surface fitt... |

1 |
polyharmonic local sine transform: A new tool for local image analysis and synthesis without edge effect
- “The
- 2006
(Show Context)
Citation Context ...m the artificial discontinuities that cause the blocking artifacts. To address these problems, Saito and Remy recently developed a new transform called Polyharmonic Local Sine Transform (PHLST) [16], =-=[17]-=-. PHLST decomposes the data in each block into two components: the polyharmonic component and the residual. The polyharmonic component is obtained by solving the polyharmonic equation (e.g., Laplace’s... |

1 |
transforms for efficient transform/subband coding
- “Lapped
- 1990
(Show Context)
Citation Context ...Remy proposed the Polyharmonic Local Sine Transform (PHLST) [16], [17]. PHLST also compensates several problems in the local trigonometric transforms (LTTs) of Coifman and Meyer [20] and Malvar [21], =-=[22]-=-, such as the overlapping windows and the slope of the bell functions. Although PHLST can be defined quite generally as in [17], we restrict ourselves to the simplest case in February 17, 2006 DRAFT 5... |