Distributions of the Two-Dimensional DCT Coefficients for Images

Reininger, R.; Gibson, J.D.

doi:10.1109/tcom.1983.1095893

Cited by 428 publications

(171 citation statements)

References 5 publications

Supporting

Mentioning

158

Contrasting

Unclassified

Order By: Relevance

“…There are 64 frequencies, and it is known ( [9]- [12]) that the histogram of coefficients corresponding to any one of the 63 AC frequencies resembles a Laplacian distribution.…”

Section: B Continuous Laplacian Distributionmentioning

confidence: 99%

Recompression of JPEG images by requantization

Bauschke

Hamilton

Macklem

et al. 2003

IEEE Trans. on Image Process.

View full text Add to dashboard Cite

Abstract-In this paper, we report a novel heuristic for requantizing JPEG images. The resulting images are generally smaller and often have improved perceptual image quality over a "blind" requantization approach, that is, one that does not consider the properties of the quantization matrices. The heuristic is supported by a detailed mathematical treatment which incorporates the wellknown Laplacian distribution of the AC discrete cosine transform (DCT) coefficients with an analysis of the error introduced by requantization. We note that the technique is applicable to any image compression method which employs discrete cosine transforms and quantization.

show abstract

“…There are 64 frequencies, and it is known ( [9]- [12]) that the histogram of coefficients corresponding to any one of the 63 AC frequencies resembles a Laplacian distribution.…”

Section: B Continuous Laplacian Distributionmentioning

confidence: 99%

Recompression of JPEG images by requantization

Bauschke

Hamilton

Macklem

et al. 2003

IEEE Trans. on Image Process.

View full text Add to dashboard Cite

show abstract

“…Real images usually significantly dither from Gaussian distribution. We use DCT coefficients as a host signal for above-mentioned reason, thus we are dealing with values that are approximated by Laplacian probability distribution functions [7].…”

Section: Resultsmentioning

confidence: 99%

Real data performance evaluation of CAISS watermarking scheme

Guzik

Matiolański

Dziech

2013

Multimed Tools Appl

View full text Add to dashboard Cite

In this paper we present real data bit error rate (BER) performance evaluation of the recently proposed correlation-and-bit-aware improved spread spectrum (CAISS) watermarking scheme (Valizadeh and Wang, IEEE Trans Inf Forensics Secur 6(2):267-282, 2011). Our tests were performed in discrete cosine transform (DCT) domain. The results show significant improvement as compared with traditional spread spectrum technique. Tests performed under medium JPEG compression and fixed peak signal-to-noise ratio (PSNR) indicate that appropriate choice of CAISS parameters allows for over a three orders of magnitude smaller BER compared to spread spectrum technique without side information about correlation. CAISS also significantly outperforms traditional spread spectrum when watermarked images are subjected to additive Gaussian noise or downscaled before watermark decoding. We also compared CAISS with the improved spread spectrum (ISS) scheme and found that CAISS can perform significantly better than ISS (in terms of BER) after medium JPEG compression but gives almost the same results in presence of additive Gaussian noise and image scaling.

show abstract

“…How ever, it has been argued that the normal distribution may not be the best model for every type of image. For example, the work [5] suggested using the Laplace distribution for certain images. Other nonnormal distributions have also been used, see, e.g., [3].…”

Section: Bessel Function G Suppose G Is the Bessel Function Pdf Givementioning

confidence: 99%