CN107609595B - Line cutting image detection method - Google Patents
Line cutting image detection method Download PDFInfo
- Publication number
- CN107609595B CN107609595B CN201710846757.6A CN201710846757A CN107609595B CN 107609595 B CN107609595 B CN 107609595B CN 201710846757 A CN201710846757 A CN 201710846757A CN 107609595 B CN107609595 B CN 107609595B
- Authority
- CN
- China
- Prior art keywords
- features
- training
- image
- wld
- lbp
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000005520 cutting process Methods 0.000 title claims abstract description 19
- 238000001514 detection method Methods 0.000 title claims abstract description 16
- 238000012549 training Methods 0.000 claims abstract description 30
- 238000012360 testing method Methods 0.000 claims abstract description 26
- 238000000034 method Methods 0.000 claims abstract description 22
- 230000004927 fusion Effects 0.000 claims abstract description 5
- 238000012545 processing Methods 0.000 claims abstract description 4
- 238000007619 statistical method Methods 0.000 claims description 4
- 238000012313 Kruskal-Wallis test Methods 0.000 claims 1
- 239000013598 vector Substances 0.000 claims 1
- 239000000284 extract Substances 0.000 description 3
- 108010068977 Golgi membrane glycoproteins Proteins 0.000 description 1
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 238000012217 deletion Methods 0.000 description 1
- 230000037430 deletion Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 238000004321 preservation Methods 0.000 description 1
- 230000035755 proliferation Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Images
Landscapes
- Image Analysis (AREA)
Abstract
The invention relates to a line cutting image detection method, which comprises two stages of training and testing; a training stage step: performing graying processing on the training set image, extracting LBP (local binary pattern) features and WLD (white line definition) features of the training set image one by one, performing feature selection on the extracted features by using a Kruskal-Wallis method, determining an optimal threshold value, and performing series fusion on the LBP features and the WLD features after feature selection to form a training feature set; a testing stage step: graying the test image according to the training stage steps, extracting LBP (local binary pattern) features and WLD (white line definition) features, performing feature selection on the extracted LBP features and WLD features by using an optimal threshold value determined in the training stage, and performing feature fusion to form a test feature set; and automatically searching an optimal parameter by the SVM classifier to judge whether the test image is subjected to line cutting operation. The method can be widely applied to the field of line cutting image detection, and the detection accuracy is improved.
Description
Technical Field
The invention relates to a line cutting image detection method.
Background
With the proliferation of inexpensive, portable image capture devices, almost everyone can record, store, and share digital images. Meanwhile, it becomes easier for a common user to tamper with the image by using different image editing software; moreover, it is very difficult or even impossible to distinguish the authenticity of the tampered image by the naked eye. Therefore, image forensics detection is a very active research direction in the field of information security. The line cutting method is an image scaling method which is widely used and keeps the content of an image unchanged; it achieves a preferential zoom performance, in particular a good compromise in zooming between the protection of important areas of the image and the preservation of the content as a whole. The line clipping method has been embedded into PhotoShop CS 6 and GIMP as an adaptive scaling method. However, line cropping methods are also used for the purpose of tampering with the image content. First, it may be used to correct a component of an image, which would be a fraud if the resulting picture was used for a competition; secondly, it may also be used intentionally to perform object deletion, which would change the semantics of the image. Therefore, it is worth studying to design a passive evidence-obtaining method to reveal those through line clipping methods.
In recent years, there have been some image detection methods related to line cropping; initially, forensic hashing was used to detect line-cut images, which effectively detected whether the image was scaled by a line-cut method; however, the forensic hash is usually established in the original image in advance, which is an active forensic method; and it may be removed by a tamperer causing a detection failure. In the blind forensics method, 324-dimensional markov features are extracted from candidate images to reveal line-cut images; the wavelet absolute moment is used for extracting the characteristics of the line cutting image; some use the packet decomposition method to extract the characteristic of the suspected image to carry on the recognition of the line cutting image; some methods extract features of the suspicious image based on energy bias and noise to identify whether the image is subjected to a line cropping operation.
Although these algorithms are relatively sophisticated, these methods all suffer from the same disadvantages. The inherent mechanism of the line clipping operation is not fully considered, nor is the visual distortion brought about by the line clipping operation further explored.
The invention content is as follows:
the invention provides a line cutting image detection method, which can detect whether images to be tested are subjected to line cutting operation in batch.
The technical scheme for solving the problems is divided into two stages, and the specific steps are as follows:
a training stage:
the method comprises the following steps: judging whether the training set image is a gray image or not; if not, graying the image;
step two: extracting LBP characteristics and WLD characteristics of the training set images one by one; LBP characteristics: using 1 as neighborhood radius, 8 neighborhoodsLBP coding with pixels representing the center pixel; thus generating a 28The WLD feature divides the image into 3 × 3 blocks each containing 8 directions, extracts an 8-dimensional histogram in each direction, then takes the entire image as one block, extracts an 8-dimensional histogram in each direction, and finally generates ((3 × 3) +1) × 8 × 8-640-dimensional features.
Step three: performing feature extraction on the features extracted in the second step by using a Kruskal-Wallis statistical method, and removing features which do not contribute to the final recognition rate;
step four: performing feature fusion on the LBP feature and the WLD feature obtained after the feature selection in the step three;
and (3) a testing stage:
the method comprises the following steps: carrying out graying processing on the test image according to the operation of the first step in the training stage;
step two: extracting LBP characteristics and WLD characteristics from the test image according to the method of the second training stage;
step three: performing feature selection operation on the extracted LBP feature and WLD feature according to a Kruskal-Wallis statistical method used in the third step of the training stage;
step four: performing fusion operation on the LBP characteristic and the WLD characteristic subjected to the characteristic selection operation;
step five: transmitting the feature set obtained in the step four and the feature set obtained in the step four in the training stage to an SVM classifier; the classifier can automatically find the optimal parameters to judge whether the image to be tested is subjected to line cutting operation.
The invention has the beneficial effects that: the invention relates to a line cutting image detection method, which comprises two stages: a training stage and a testing stage; and (3) carrying out feature selection on the features extracted in the training stage by a Kruskal-Wallis statistical method to form a training feature set, extracting a test feature set in the testing stage, transmitting the training feature set and the test feature set to an SVM classifier, and finally automatically judging whether the test image is subjected to line cutting operation by the classifier. The method fully considers the influence of line cutting on the image and improves the detection accuracy. Therefore, the method can be widely applied to the field of line cutting image detection.
Drawings
Fig. 1 is a general flowchart of a line-cutting image detection method according to the present invention.
Claims (2)
1. A line cutting image detection method is characterized by comprising the following specific steps:
a training stage:
a training stage comprises the following steps: judging whether the training set image is a gray image, if not, performing graying processing on the training set image;
a second step in the training stage: extracting LBP features and WLD features of the training set images one by one; LBP characteristics: taking 1 as a neighborhood radius, and expressing LBP coding of a central pixel by 8 neighborhood pixels; thus generating a 28The WLD feature is that the image is divided into 3 × 3 blocks, each block comprises 8 directions, 8-dimensional histograms are extracted in each direction, then the whole image is taken as one block, 8-dimensional histograms are extracted in each direction, and finally ((3 × 3) +1) × 8 × 8 is generated as 640-dimensional features;
a training stage comprises the following steps: respectively selecting the extracted LBP characteristics and WLD characteristics by using a Kruskal-Wallis statistical method; determining an optimal threshold value through a Kruskal-Wallis test, wherein the features smaller than the threshold value are left, and the features larger than the threshold value are automatically deleted;
step four: carrying out series fusion on the LBP characteristic and the WLD characteristic after characteristic selection to form a training characteristic set;
and (3) a testing stage:
the first step of the test stage: carrying out graying processing on the test image according to the first step of the training stage;
a second testing stage step: extracting LBP characteristics and WLD characteristics of the test image;
a third step of the test stage: performing feature selection operation on the LBP features and the WLD features extracted in the second step of the testing stage by using an optimal threshold determined by Kruskal-Wallis testing in the training stage, and removing features which do not contribute much to the recognition rate; fusing the LBP characteristic and the WLD characteristic after the characteristic selection operation to form a test characteristic set;
the fourth step of the test stage: transmitting the training feature set obtained in the fourth training stage step and the test feature set obtained in the third testing stage step to an SVM classifier as feature vectors; the SVM classifier automatically searches for optimal parameters to classify the feature set, so that whether the test image is subjected to line cutting operation or not is determined.
2. The line cropping image detection method of claim 1, wherein the extracted LBP feature and WLD feature are respectively subjected to feature dimension reduction by using Kruskal-Wallis method, so as to obtain an optimal threshold; this threshold is automatically optimized by the Kruskal-Wallis method, rather than being manually selected.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710846757.6A CN107609595B (en) | 2017-09-19 | 2017-09-19 | Line cutting image detection method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710846757.6A CN107609595B (en) | 2017-09-19 | 2017-09-19 | Line cutting image detection method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107609595A CN107609595A (en) | 2018-01-19 |
CN107609595B true CN107609595B (en) | 2020-09-11 |
Family
ID=61061112
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710846757.6A Active CN107609595B (en) | 2017-09-19 | 2017-09-19 | Line cutting image detection method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107609595B (en) |
Families Citing this family (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110021036B (en) * | 2019-04-13 | 2021-03-16 | 北京环境特性研究所 | Infrared target detection method and device, computer equipment and storage medium |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105488758A (en) * | 2015-11-30 | 2016-04-13 | 河北工业大学 | Image scaling method based on content awareness |
CN105825183A (en) * | 2016-03-14 | 2016-08-03 | 合肥工业大学 | Face expression identification method based on partially shielded image |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US9443169B2 (en) * | 2014-02-21 | 2016-09-13 | Xerox Corporation | Object classification with constrained multiple instance support vector machine |
-
2017
- 2017-09-19 CN CN201710846757.6A patent/CN107609595B/en active Active
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105488758A (en) * | 2015-11-30 | 2016-04-13 | 河北工业大学 | Image scaling method based on content awareness |
CN105825183A (en) * | 2016-03-14 | 2016-08-03 | 合肥工业大学 | Face expression identification method based on partially shielded image |
Non-Patent Citations (3)
Title |
---|
基于局部二元模式的快速红外人脸识别系统;谢志华;《红外与激光工程》;20131225;第42卷(第12期);第3191-3193页第1-3节 * |
数字图像盲鉴别的关键理论与技术研究;吕颖达;《中国博士学位论文全文数据库 信息科技辑》;中国学术期刊(光盘版)电子杂志社;20150815(第08期);第100-101页,附图5.1 * |
杨富圣 等.—种基于LBP特征的Seam-Carving图像篡改检测方法.《南开大学学报(自然科学版)》.2015,第48卷(第02期),第76-78页. * |
Also Published As
Publication number | Publication date |
---|---|
CN107609595A (en) | 2018-01-19 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Nishiyama et al. | Facial deblur inference using subspace analysis for recognition of blurred faces | |
Jaiswal et al. | Image splicing detection using deep residual network | |
CN106060568B (en) | A kind of video tampering detection and localization method | |
CN114897806A (en) | Defect detection method, electronic device and computer readable storage medium | |
Zhang et al. | Detection of LSB matching steganography based on distribution of pixel differences in natural images | |
Rhee | Detection of spliced image forensics using texture analysis of median filter residual | |
Barni et al. | Detection of adaptive histogram equalization robust against JPEG compression | |
Sebastian et al. | Image authentication by content preserving robust image hashing using local and global features | |
Steinebach et al. | Efficient cropping-resistant robust image hashing | |
Chetty et al. | Nonintrusive image tamper detection based on fuzzy fusion | |
CN111259792A (en) | Face living body detection method based on DWT-LBP-DCT characteristics | |
Celiktutan et al. | Blind identification of cellular phone cameras | |
CN107609595B (en) | Line cutting image detection method | |
Vaishnavi et al. | Recognizing image splicing forgeries using histogram features | |
Tripathi et al. | Automated image splicing detection using texture based feature criterion and fuzzy support vector machine based classifier | |
Mishra et al. | Detection of clones in digital images | |
Lin et al. | Face detection based on skin color segmentation and SVM classification | |
KR20060007901A (en) | Apparatus and method for automatic extraction of salient object from an image | |
Ahmed et al. | Image splicing detection and localisation using efficientnet and modified u-net architecture | |
Patil et al. | Image hashing by SDQ-CSLBP | |
Vashishtha et al. | Least significant bit matching steganalysis based on feature analysis | |
Gupta et al. | Energy deviation measure: a technique for digital image forensics | |
Fan et al. | Image tampering detection using noise histogram features | |
Essa et al. | High Order Volumetric Directional Pattern for Video‐Based Face Recognition | |
CN108737814B (en) | Video shot detection method based on dynamic mode decomposition |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant | ||
TR01 | Transfer of patent right |
Effective date of registration: 20231107 Address after: 518000 District D and E, 7th Floor, Building 3, Tingwei Industrial Park, 6 Liufang Road, Xin'an Street, Baoan District, Shenzhen City, Guangdong Province Patentee after: Shenzhen Aixiesheng Technology Co.,Ltd. Address before: 410114 No. 960, Section 2, Wanjiali South Road, Tianxin District, Changsha City, Hunan Province Patentee before: CHANGSHA University OF SCIENCE AND TECHNOLOGY |
|
TR01 | Transfer of patent right |