A method and system for detecting length and location of payload embedded in an image
Patent Information
- Application Number
- IN202211003092
- Authority / Receiving Office
- IN · IN
- Patent Type
- Patents
- Current Assignee / Owner
- Filing Date
- 2022-01-19
- Publication Date
- 2026-08-12
- Estimated Expiration
- 2042-01-19
AI Technical Summary
Current steganalysis methods face challenges in efficiently detecting stego images with low volume payloads and identifying the specific steganographic tools used, particularly due to high computation time and limited accuracy in payload length calculation.
A method and system employing multiple color models and a fluctuation function to classify stego images by extracting feature vectors from split color channels, segmenting into non-overlapping blocks, and using a trained classifier to identify payload locations and calculate hidden message length.
This approach improves classification accuracy and reduces computation time by effectively identifying steganographic tools and estimating payload length in images with low volume payloads, enhancing the detection of stego images and their embedded information.
Abstract
Description
TECHNICAL FIELD:
[0001] The present subject matter described herein, relates to a method and system for detection of spatial steganographic tools to classify stego images created with a particular load and low volume payloads for calculating length of the hidden message in a given stego image. The present invention, in particular, deals with active steganalysis of steganographic tools to classify stego images created with a particular tool and low volume payloads and thereby calculating length of the hidden message in a given stego image.BACKGROUND OF THE INVENTION:
[0002] Background description includes information that may be useful in understanding the present invention.
[0003] With the advent of technology and social media, more and more people share images as a part of daily routine. This huge volume of images shared may serve as an abundant source for the illicit users to exploit them to transmit stolen / malicious information innocuously. Steganalysis is the process of finding out the embedding information covertly hidden inside images. It is an inverse process of steganography. Steganography is the technique that hides information inside the images as noise.
[0004] Steganalysis is an art of deterring covert communications while allowing the innocent ones. It can be either passive or active. In passive steganalysis, the aim is to identify whether an image carries a payload or not, whereas in active steganalysis the specific algorithm / tool which is used to embed the payload has to be identified. Over the past two decades, many approaches of steganalysis have been proposed. Traditionally, those methods try to resolve the problem using two stages. The first stage is extraction of handcrafted features which must be able to model the embedding distortions caused by any steganographic algorithm in the image. The second stage is classification using any binary classifier to simply identify the presence of secret information or ensemble of classifiers to predict the steganographic algorithms / tools used to embed the secret. These traditional techniques fall under machine learning domain because, after the feature extraction process is carried out with human supervision, the machine (computer) is trained in such a way that it automatically learns to separate different classes from the training data.
[0005] The most popular hand-crafted features are the Spatial Rich Model (SRM) features. In SRM feature extraction framework, an image is firstly filtered by a set of high-pass filters to generate residual images. Then, joint distribution models of quantized residuals are obtained by computing the fourth order co-occurrence matrices. Finally, steganalytic features are obtained by merging elements in the co-occurrence matrices. A set of linear and non-linear high pass filters with different dimensions and orientations, together with several residual quantization steps, are applied to increase model diversity. Therefore, intricate high-order relationships among pixels can be modelled to quantify steganographic anomalies.
[0006] In steganalysis, texture features play a major role since the adaptive steganographic algorithms mostly restrain to embedding in these regions. Such a kind of texture feature is Local Binary Pattern (LBP) operator. A set of high-order derivative filters is used to obtain residual images and then the Threshold LBP (TLBP) operation is performed on them. From the TLBP images, second order co-occurrence matrix features are formed by feature aggregation process similar to the co-occurrence symmetrization in SRM.
[0007] The frame-based steganalysis, which extracts a set of features from the whole image is most commonly found in literature. In contrast, the block-based steganalysis which extracts features from each individual block is very rare. Based on the block features, a Tree-Structured Vector Quantization (TSVQ) scheme has been adopted to classify blocks into multiple classes. For each class, a specific classifier can be trained using block features, which represents the characteristics of the block class. For a given test image, instead of making a single decision for the entire image, the block decomposition process is repeated and a classifier is chosen to make a cover / stego decision for each block depending on block features. Finally, a decision fusion technique can be used to fuse steganalysis results of all blocks so that one can decide whether an unknown image is a cover or stego image.
[0008] The aforementioned steganalysis methods follow a two-stage detection process i.e. hand crafted feature extraction and classification. This is replaced by a single stage detection process by Conventional Neural Network (CNN) based steganalysis models. Recently, a lot of work is going on in steganalysis based on CNNs. In the first deep learning model for steganalysis, the implementation of an unsupervised pre-training procedure using Stacked Convolutional Auto Encoders (SCAE) is proposed. This model consists of multiple convolution and max pooling layers with different number of kernels. Then binary classifier (stego or cover) is implemented by two fully connected layers.
[0009] Even though CNN based steganalysis methods have achieved remarkable performance, its high computation time makes its practical implementation difficult. Also, low volume payload detection task is challenging.
[0010] Accordingly, there is a need for a method and a system for steganalysis of an input image by employing multiple colour models simultaneously to improve classification rate.
[0011] In view of the above, there is requirement to devise an apparatus which can address the aforesaid shortcoming(s). Hence, the present invention has been introduced to overcome this problem.
[0012] The information disclosed in this background of the subject matter section is only for enhancement of understanding of the general background of the invention and should not be taken as an acknowledgement or any form of suggestion that this information forms the prior art already known to a person skilled in the art.OBJECTS OF THE INVENTION:
[0013] It is therefore an object of the present subject matter to overcome the aforementioned and other drawbacks in the solutions available in state-of-the-art.
[0014] The principal objective of the present subject matter is to provide a method and system to steganalysis of steganographic tools to classify stego images created with a particular tool and low volume payloads for calculating length of the hidden message in a given stego image.
[0015] Said object is achieved according to the invention by the features of the independent claim 1. Further, particularly advantageous embodiments of the present subject matter are disclosed in the sub-claims.
[0016] Another object of the present subject matter is to provide a system to perform active steganalysis of steganographic tools when images are embedded with minute payloads.
[0017] A yet another object of this invention is to provide weak embedding distortions by a discriminative feature, 'fluctuation function'.
[0018] Still another object of this invention is to identify the tool employed in the steganographic process by means of fluctuation function and a trained classifier.
[0019] Further object of this invention is to locate the payload bits and calculate its length from the estimated locations.
[0020] These and other objects and advantages of the present subject matter will be apparent to a person skilled in the art after consideration of the following detailed description taken into consideration with accompanying drawings in which preferred embodiments of the present subject matter are illustrated.SUMMARY OF THE INVENTION:
[0021] This summary is provided to introduce concepts related to a system and a method for steganalysis of an input image. The concepts are further described below in the detailed description. This summary is not intended to identify key features or essential features of the claimed subject matter, nor is it intended to be used to limit the scope of the claimed subject matter.
[0022] The present subject matter described herein, relates to a method and system for detection of spatial steganographic tools to classify stego images created with a particular load and low volume payloads for calculating length of the hidden message in a given stego image. The method comprises classifying stego images by identifying payload embedded and a steganographic tool employed in an arbitrary input image, locating payload embedded locations in the stego image, and calculating the length of the hidden message in the stego image using payload locations and tool employed in the stego image.
[0023] In an aspect, classifying stego images comprising: performing feature extraction on a given stego image from the training images dataset containing cover and stego images of various steganographic tools to determine a feature vector of the given stego image, using feature vectors determined from the extracted features, to train a classifier model, performing feature extraction on an arbitrary image to determine the feature vector of the arbitrary image, and using trained classifier to classify stego images from the arbitrary input images.
[0024] In an aspect, the feature extraction on the image comprises: splitting the obtained cross colour residue image into three separate channel; extracting the Least Significant Bit plane of each channel of the cross colour residue image; segmenting the Least Significant Bit plane into N number of non-overlapping block; determining a fluctuation feature vector for each of the three separated channel; and determining feature vectors of the given stego image using fluctuation features of each of the three separated channel.
[0025] In an aspect, the number of non-overlapping blocks are determined by the equation:N= mxn / rxsWherein rxs is size of each non-overlapping blockmxn is channel size.
[0026] In as aspect, the fluctuation feature of each of the three separated channel is determined by the equation:EquationWherein xixi,j,j is a pixel in a block(i, j) indicates the pixel position(fluct) is the fluctuation function
[0027] The present subject matter further relates to a system for detection of spatial steganographic tools to classify stego images created with a particular load and low volume payloads for calculating length of the hidden message in a given stego image. The system comprises a display, a memory and a processor. The processor is operatively coupled with the memory. The processor configured to classify the stego images by identifying payload embedded and a steganographic tool employed in an arbitrary input image, locating the payload embedded locations in the stego image and calculating length of the hidden message in the stego image using payload locations and tool employed in the stego image.
[0028] In an aspect, the processor is further configured to classify the stego images by: performing the feature extraction on a given stego image from the training images dataset containing cover and stego images of various steganographic tools to determine a feature vector of the given stego image, using feature vectors determined from the extracted features, to train a classifier model, performing feature extraction on an arbitrary image to determine the feature vector of the arbitrary image and using trained classifier to classify stego images from the arbitrary input images.
[0029] In an aspect, the processor is further configured for feature extraction on the image by: splitting the obtained cross colour residue image into three separate channel; extracting the Least Significant Bit plane of each channel of the cross colour residue image; segmenting the Least Significant Bit plane into N number of non-overlapping block; determining the fluctuation feature for each of the three separated channel; and determining feature vectors of the image using fluctuation features of each of the three separated channel.
[0030] To further understand the characteristics and technical contents of the present subject matter, a description relating thereto will be made with reference to the accompanying drawings. However, the drawings are illustrative only but not used to limit the scope of the present subject matter.
[0031] Various objects, features, aspects, and advantages of the inventive subject matter will become more apparent from the following detailed description of preferred embodiments, along with the accompanying drawing figures in which like numerals represent like components.BRIEF DESCRIPTION OF FIGURES
[0032] The illustrated embodiments of the present disclosure will be best understood by reference to the drawings, wherein like parts are designated by like numerals throughout. The following description is intended only by way of example, and simply illustrates certain selected embodiments of devices, systems, and processes that are consistent with the subject matter as claimed herein, wherein:
[0033] FIG. 1 illustrates an exemplary block chart of a method for detection of spatial steganographic tools to classify stego images created with a particular load and low volume payloads for calculating length of the hidden message in a given stego image that can be utilized to implement one or more exemplary embodiments of the present disclosure;
[0034] FIG. 2 illustrates a schematic diagram for steganalysis based on fluctuation feature that can be utilized to implement one or more exemplary embodiments of the present disclosure;
[0035] FIG. 3 illustrates a block diagram representing the procedure for location and length estimation of payload, that can be utilized to implement one or more exemplary embodiments of the present disclosure;
[0036] FIG. 4 illustrates a block diagram of visual interpretation of location and length estimation of payload that can be utilized to implement one or more exemplary embodiments of the present disclosure; and
[0037] FIG. 5 illustrates a system for detection of spatial steganographic tools to classify stego images created with a particular load and low volume payloads for calculating length of the hidden message in a given stego that can be utilized to implement one or more exemplary embodiments of the present disclosure.
[0038] The figures depict embodiments of the present subject matter for the purposes of illustration only. A person skilled in the art will easily recognize from the following description that alternative embodiments of the structures and methods illustrated herein may be employed without departing from the principles of the disclosure described herein.DETAILED DESCRIPTION
[0039] A few aspects of the present disclosure are explained in detail below with reference to the various figures. Example implementations are described to illustrate the disclosed subject matter, not to limit its scope, which is defined by the claims. Those of ordinary skill in the art will recognize a number of equivalent variations of the various features provided in the description that follows.
[0040] The present subject matter described herein, relates to a method and system for detection of spatial steganographic tools to classify stego images created with a particular load and low volume payloads for calculating length of the hidden message in a given stego image.
[0041] Some embodiments of this invention, illustrating all its features, will be discussed in detail.
[0042] FIG. 1 illustrates an exemplary block chart of a method 100 detection of spatial steganographic tools to classify stego images created with a particular load and low volume payloads for calculating length of the hidden message in a given stego image that can be utilized to implement one or more exemplary embodiments of the present disclosure.
[0043] At block 101, an image which is to be scrutinized is selected as input image. For purpose of clear understanding of the subject matter, it is assumed that the input image is an arbitrary image containing embedded information.
[0044] At block 103, the input image is processed to classify stego images by identifying payload embedded and a steganographic tool employed in an arbitrary input image.
[0045] At block 104, detection of payload embedded locations in performed.
[0046] At block 105, based on the payload embedded locations, calculation of the length of the hidden message is performed.
[0047] FIG. 2 illustrates the schematic of performing feature extraction for the proposed invention. An input image typically consists of heterogeneous regions. In this invention, an image is split into smaller blocks and each block is used as a basic unit for steganalysis. The stego images in the steganographic tools database available are in colour RGB format i.e. Red, Green and Blue channels. The first step is to split the image into three separate channels. Then the Least Significant Bit (LSB) plane of each channel is extracted and then this plane is segmented into N number of non-overlapping blocks, each of size r x s. Assuming the channel size as m x n, the number of such non-overlapping blocks, is given by the Equation (1):N = mxn / rxsm X nr X s ……………………… (1)Wherein rxs is size of each non-overlapping blockmxn is channel size.
[0048] Considering a pixel xij in a block, where (i, j) indicates the pixel position, the fluctuation function (fluct) to measure the fluctuation of each block is given in Equation (2):Equation
[0049] Since this feature extracts a cumulative sum of differences between a pixel's LSB and its neighbors within the block, the minute changes in the bits reflect the embedding pattern of specific tools. For all the three colour planes, the feature vectors are extracted. From a training images dataset containing cover and stego images of various steganographic tools, feature vectors are derived and used to train a classifier model. The trained classifiers are deployed in a hardware system for testing unseen test images.
[0050] FIG. 3 & 4 illustrate the procedure to estimate the length and location of payload hidden in the given stego image. After identifying the specific stego tool from an input image, cover estimation from the stego image need to be performed. To achieve this purpose, stego image is considered to be a cover image which is corrupted by stego noise. Hence, using any image denoising method, stego image is denoised so that stego noise is removed from it to obtain an estimate of cover. From this estimated cover image, fluctuation feature is extracted. Absolute difference between the fluctuation function of stego and estimated cover is calculated. In this difference feature, values of fluctuation function that are equal to zero, correspond to the regions that are same in stego and cover. Thus, the non-zero values in the difference feature lead to the regions where embedding has been done. Therefore, number of non-zero values in the difference feature gives an approximate length of the payload. Since fluctuation function is block based, the non-zero values in the difference feature corresponds to the location of payload blocks. The indices in stego image corresponding to the blocks with non-zero values point to the approximate payload location. By extracting the LSB of the pixel values in these locations, some approximate secret also can be extracted.
[0051] FIG. illustrates a system 500 for steganalysis of an input image that can be utilized to implement one or more exemplary embodiments of the present disclosure. The system 500 mainly comprises a display 501, a memory 502 and a processor 503. The processor 503 is operatively connected to the memory 502 and the display 501. The processor 403 is configured to execute steps 103, 104, and 105 of method 100 as already discussed in FIG. 1. Further, the processor 403 is configured to obtain fluctuation features as discussed in FIG. 2 Additionally, the processor 503 is configured to determine the difference between fluctuation features of stego and estimated cover images as discussed in FIG. 3 and FIG. 4.
[0052] In an embodiment, the steganalysis system 500 comprises of a hand held device equipped with Raspberry Pi board and standard input / output.
[0053] In experimental embodiment, a steganalysis system in accordance with the present invention has a 3 x 3 LSB block from an image as follows:1 0 00 1 10 1 0
[0054] The fluctuation function for this block is computed using Equation (1) as follows isEquationwhere, r = 3 and s =3.
[0055] Thus, with the implementation of the method 100 and the system 500 of the present subject matter, passive steganalysis could be performed. Further, the present method 100 and system 500 could be used to circumvent the blind steganalysis problem by allowing simultaneous representation of the image in hand by different colour models. Furthermore, the present method 100 and system 500 addresses the problem of null reference with steganalysis by coherently representig the image with various colour models.
[0056] The techniques described above may be implemented using one or more computer programs executing on (or executable by) a programmable computer including any combination of any number of the following: a processor, a sensor, a storage medium readable and / or writable by the processor (including for example volatile and non-volatile memory and / or storage elements), plurality of inputs units, plurality of output devices and networking devices.
[0057] Further, with the implementation of the present disclosure, thermal model based protection strategy which require additional space, additional sensors, and major modification in control unit and calibration efforts could be replaced. Thus, the present disclosure provides a better and economical replacement of thermal model based protection strategy for manual transmission vehicles.
[0058] The above description does not provide specific details of the manufacture or design of the various components. Those of skill in the art are familiar with such details, and unless departures from those techniques are set out, techniques, known, related art or later developed designs and materials should be employed. Those in the art are capable of choosing suitable manufacturing and design details.
[0059] It should be understood, however, that all of these and similar terms are to be associated with the appropriate physical quantities and are merely convenient labels applied to these quantities. Unless specifically stated otherwise, as apparent from the discussion herein, it is appreciated that throughout the description, discussions utilizing terms such as "receiving," or "retrieving," or "comparing," or "generating," or the like, refer to the action and processes of a computer system, or similar electronic computing device, that manipulates and transforms data represented as physical (electronic) quantities within the computer system's registers and memories into other data similarly represented as physical quantities within the computer system memories or registers or other such information storage, transmission or display devices.
[0060] The exemplary embodiment also relates to a system for performing the operations discussed herein. This system may be specially constructed for the required purposes, or it may comprise a general-purpose computer selectively activated or reconfigured by a computer program stored in the computer. Such a computer program may be stored in a computer-readable storage medium, such as, but is not limited to, any type of disk including floppy disks, optical disks, CD-ROMs, and magnetic-optical disks, read-only memories (ROMs), random access memories (RAMs), EPROMs, EEPROMs, solid state drives, magnetic or optical cards, or any type of media suitable for storing electronic instructions, and each coupled to a computer system bus.
[0061] Any content disclosed herein may be implemented, for example, in one or more content structures tangibly stored on a non-transitory computer-readable medium. Embodiments of the invention may store such content in such content structure(s) and read such content from such content structure(s).
[0062] Further, the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the disclosure. It will be appreciated that several of the above-disclosed and other features and functions, or alternatives thereof, may be combined into other systems or applications. Various presently unforeseen or unanticipated alternatives, modifications, variations, or improvements therein may subsequently be made by those skilled in the art without departing from the scope of the present disclosure as encompassed by the following claims.
[0063] The claims, as originally presented and as they may be amended, encompass variations, alternatives, modifications, improvements, equivalents, and substantial equivalents of the embodiments and teachings disclosed herein, including those that are presently unforeseen or unappreciated, and that, for example, may arise from applicants / patentees and others.
[0064] It will be appreciated that variants of the above-disclosed and other features and functions, or alternatives thereof, may be combined into many other different systems or applications. Various presently unforeseen or unanticipated alternatives, modifications, variations, or improvements therein may be subsequently made by those skilled in the art which are also intended to be encompassed by the following claims.
Claims
1. A method (100) for detection of spatial steganographic tools to classify stego images created with a particular load and low volume payloads for calculating length of the hidden message in an arbitrary image (101), the method (100) comprising: classifying (103), stego images by identifying payload embedded and a steganographic tool employed in an arbitrary input image; locating (104), payload embedded locations in the stego image; and calculating (105), length of the hidden message in the stego image using payload locations and tool employed in the stego image.
2. The method (100) as claimed in claim 1, wherein classifying stego images, comprising: performing, feature extraction on a given stego image (102) from the training images dataset containing cover and stego images of various steganographic tools to determine a feature vector of the given stego image; using, feature vectors determined from the extracted features, to train a classifier model (102); performing, feature extraction on an arbitrary image (101) to determine the feature vector of the arbitrary image; and using, trained classifier (102) to classify stego images from the arbitrary input images.
3. The method (100) as claimed in claim 2, wherein performing feature extraction on an image comprising: splitting, the image into three separate channels i.e. Red, Green and Blue channels; extracting, the Least Significant Bit (LSB) plane of each channel of the given stego image; segmenting, the Least Significant Bit (LSB) plane into N number of non-overlapping blocks; determining, a fluctuation feature of each non-overlapping block; and determining, feature vectors of the image using fluctuation features of each non-overlapping block.
4. The method (100) as claimed in claim 3, wherein the number of non-overlapping blocks are determined by the equation: N=mxn / rxs Wherein rxs is size of each non-overlapping block mxn is channel size.
5. The method (100) as claimed in claim 3, wherein determining, the fluctuation feature of each non-overlapping block by the equation: Equation wherein xij is a pixel in a block, where (i, j) indicates the pixel position (fluct) is the fluctuation function6. The method (100) as claimed in claim 2, wherein trained classifier (102) is used to classify the stego images on the basis of the specific class of the steganographic tool employed in the arbitrary input image (101).
7. A system (500) for steganalysis of an input image, the system comprises: a display (401); a memory (402); and a processor (403) operatively coupled with the memory (402), the processor (403) is configured to: classify the stego images by identifying payload embedded and a steganographic tool employed in an arbitrary input image (101); locating the payload embedded locations in the stego image; and calculating length of the hidden message in the stego image using payload locations and tool employed in the stego image.
8. The system (500) as claimed in claim 9, wherein the processor is further configured to classify the stego images by: performing, feature extraction on a given stego image from the training images dataset containing cover and stego images of various steganographic tools to determine a feature vector of the given stego image; using, feature vectors determined from the extracted features, to train a classifier model (102); performing, feature extraction on an arbitrary image to determine the feature vector of the arbitrary image (101); and using, trained classifier (102) to classify stego images from the arbitrary input images (101).
9. The system (500) as claimed in claim 9, wherein the processor (403) is further configured for performing feature extraction on the image by: splitting, the image into three separate channels i.e. Red, Green and Blue channels; extracting, the Least Significant Bit (LSB) plane of each channel of the given stego image; segmenting, the Least Significant Bit (LSB) plane into N number of non-overlapping blocks; determining, a fluctuation feature of each non-overlapping block; and determining, feature vectors of the given stego image using fluctuation features of each non-overlapping block.
10. The system (500) as claimed in claim 9, wherein the number of non-overlapping blocks are determined by the equation: N=mxn / rxs Wherein rxs is size of each non-overlapping block mxn is channel size.
11. The system (500) as claimed in claim 9, wherein determining, the fluctuation feature of each non-overlapping block by the equation: Equation wherein xij is a pixel in a block, where (i, j) indicates the pixel position (fluct) is the fluctuation function12. The system (500) as claimed in claim 8, wherein trained classifier (102) is used to classify the stego images on the basis of the specific class of the steganographic tool employed in the arbitrary input image (101).
