CN105049867A

CN105049867A - Method for recognizing whether bitmap having been compressed by JPEG

Info

Publication number: CN105049867A
Application number: CN201510522016.3A
Authority: CN
Inventors: 张军; 袁志斌; 王志坚; 沈永珞
Original assignee: Guangdong Finance And Economics University
Current assignee: Guangdong Finance And Economics University; Guangdong University of Business Studies
Priority date: 2015-08-19
Filing date: 2015-08-19
Publication date: 2015-11-11

Abstract

The invention relates to a method for recognizing whether a bitmap has been compressed by JPEG and belongs to the field of multimedia information safety. In terms of the problem whether one bitmap has been compressed by JPEG, the technical proposal of the invention is as follows: performing 8 plus 8 blocking to the bitmap to be detected, and calculating DCT coefficient matrix to each block; then intercepting the fractional parts of all DCT coefficients; then calculating variances of these fractional parts, and judging that the detected bitmap has not been compressed by JPEG if the magnitude of the variance is more than a preselected threshold value, and judging that the detected bitmap has been compressed by JPEG if the magnitude of the variance is less than the preselected threshold value. The method of the invention could be used for accurately judging whether the bitmap has been compressed by the JPEG by analyzing the fractional parts of the DCT coefficients, and is low in computation complexity and strong in resolving ability. The method of the invention is conducive to tampering a follow-up detection image and hiding information.

Description

A kind ofly identify bitmap whether through the method for JPEG compression

Technical field

The invention belongs to field of multi-media information safety, particularly relate to a kind of method whether bitmap compresses through JPEG that identifies.

Background technology

The content of bitmap is easy to be modified, therefore just have be maliciously tampered and for illegal object may.In addition, bitmap is easily hidden information, for confidential corespondence, thus jeopardizes information security.Identify whether bitmap can contribute to distorting and Information hiding of subsequent detection image through JPEG compression.

Whether Artificial Cognition's bitmap compresses poor accuracy and inefficiency through JPEG, and therefore feasible scheme develops effective computer automatic identification method exactly.

JPEG is the lossy compression of extensive use in image capture apparatus, its operation principle is: JPEG compression is the image compression scheme based on piecemeal, first Iamge Segmentation is become the 8x8 piecemeal of non-overlapping copies, then the DCT coefficient matrix of 8x8 is obtained after discrete cosine transform (DCT) being carried out to each piecemeal, carry out quantification according to the quantization table of a 8x8 to DCT coefficient matrix again to round, finally obtained coefficient is encoded, generate jpeg file.First read quantization table from jpeg file head during jpeg decompression, and file code stream is decoded, obtain quantization DCT coefficient.According to quantization table, inverse quantization, reverse discrete cosine transform (IDCT) are carried out to quantization DCT coefficient again, and round, finally restore bitmap.

The Chinese patent of prior art 1: patent No. CN103067713A: a kind of method and system of bitmap JPEG compressed detected propose and judge by factor histogram the method whether bitmap compresses through JPEG, its basic process is as follows: carry out piecemeal to data bitmap to be detected and calculate corresponding DCT coefficient matrix, be corresponding integer matrix by described DCT coefficient matrix conversion, draw a factor histogram representing the number of times that each positive factor occurs respectively in described integer matrix; Analyze the histogrammic monotonicity of described factor and obtain a metric; Judging whether described metric is greater than previously selected threshold values, when being, then judging that described bitmap to be detected compresses through JPEG, when no, judge that described bitmap to be detected compresses without JPEG.The calculating more complicated of the method, the bitmap resolving ability compressed to high-quality is limited, and therefore prior art need to improve and development.

In JPEG operation principle, DCT coefficient experienced by quantification floor operation, and this operation causes the DCT coefficient of inverse quantization to be integer.Therefore, the decimal characteristic of DCT coefficient analyzing bitmap identifies piece image whether through the good method of overcompression.

Summary of the invention

The invention provides a kind of detection method whether bitmap compresses through JPEG that identifies.In theory, the fractional part without its piecemeal DCT coefficient of bitmap of compression obeys being uniformly distributed between-0.5 to+0.5; And its piecemeal DCT coefficient of compressed bitmap should be integer, but owing to introducing approximation error when being decompressed to pixel, this error approximate normal distribution obeyed between-0.5 to+0.5 after DCT linear combination of 8x8 pixel, therefore its piecemeal DCT coefficient of compressed bitmap is no longer integer, and its fractional part is similar to the normal distribution obeyed between-0.5 to+0.5.The present invention has caught the distribution characteristics of DCT coefficient fractional part, provides a kind of simple, the recognition methods that verification and measurement ratio is high.The method invented is: carry out 8x8 piecemeal to data bitmap to be detected, corresponding DCT coefficient matrix is calculated to every block, then intercept so the fractional part of DCT coefficient, calculate the variance of fractional part again, if the value of variance is greater than previously selected threshold values, then judge that the bitmap detected compresses without JPEG; If be less than previously selected threshold values, then judge that the bitmap detected have passed through JPEG compression.The effect of the method is in table 1:

Table 1

	256x256	128x128	64x64	32x32	16x16	8x8
							Prior art 1	99.32	98.70	97.34	94.86	88.02	78.44
The present invention	99.85	99.90	99.85	99.85	99.10	89.00

As can be seen from Table 1, the present invention is higher than the recognition accuracy of prior art 1, and when being greater than 16x16 to picture size, recognition accuracy of the present invention is all more than 99%.

Embodiment

Iamge Segmentation is become the 8x8 piecemeal of non-overlapping copies, then corresponding DCT coefficient matrix is calculated to every block.

Intercept the fractional part of all DCT coefficient, then calculate the variance of these fractional parts.

If the value of variance is greater than previously selected threshold values, then judge that the bitmap detected compresses without JPEG; If be less than previously selected threshold values, then judge that the bitmap detected have passed through JPEG compression.

The selecting step of above-mentioned threshold values is: the bitmap of optional some uncompressed, calculates respective corresponding variance in a manner described, add up the average of these variances, be designated as a every width bitmap; Meanwhile, compress the JPEG of above-mentioned bitmap through the various factor, then decompress(ion) is reduced to bitmap, then to the bitmap after these decompress(ion)s, calculates respective corresponding variance in a manner described, add up the average of these variances, be designated as b; Finally choose (a+b)/2 for threshold value.

Claims

1. identify and it is characterized in that the method whether bitmap compresses through JPEG said method comprising the steps of:

A. adopt the fractional part of bitmap piecemeal DCT coefficient for analyzing foundation;

B. with the variance of fractional part for criterion of identification;

If C. the value of variance is greater than previously selected threshold values, then judge that bitmap compresses without JPEG; Otherwise this bitmap have passed through JPEG compression.

2. method according to claim 1, is characterized in that the determining step of described threshold values is:

D. the bitmap of optional some uncompressed, calculates respective corresponding variance, adds up the average of these variances, be designated as a;

E. compress the JPEG of above-mentioned bitmap through the various factor, then decompress(ion) is reduced to bitmap, finally calculates respective corresponding variance to the bitmap after these decompress(ion)s, adds up the average of these variances, be designated as b;

F. (a+b)/2 are chosen for threshold value.