ROC curve for performance comparison

ROC curve for performance comparison

Source publication
Article
Full-text available
Perceptual image hashing technique uses the appearance of the digital media object as human eye and generates a fixed size hash value. This hash value works as digital signature for the media object and it is robust against various digital manipulation done on the media object. This technique have been constantly in use in various application areas...

Similar publications

Article
Full-text available
This paper employed an Interval Type-2 Fuzzy Logic (IT2FL) in the detection of fire outbreak using environmental parameters. Fire outbreak data capturing device (FODCD) is developed and used in capturing environmental parameter values. The FODCD is comprised of DHT11 temperature sensor, MQ-2 smoke sensor, LM393 Flame sensor, and ESP8266 Wi-Fi modul...

Citations

... Recently, the use of hash functions in multimedia applications for both indexing and security has caught everyone's attention. Perceptual hash functions bagged its place in several application areas, namely, tamper detection [30,35], authentication of images [3], digital watermarking [8], retrieval of images [28], copy detection in images [31], indexing of images [12], image similarity and discrimination [9], etc. The properties of a hash function may certainly vary with the type of application, but in order to be good, a hash function should have the following two aspects: ...
... The method shows robust results against various content preserving attacks, e.g., watermark embedding, additive noise, etc., but doesn't deliver good results against geometric attacks which are solved in [35]. Gharde et al. [9] proposed a perceptually robust hash function using fuzzy color histogram. The unbiased fuzzy histogram is then standardized to make it scale-invariant and finally, hashes are calculated from it. ...
... For evaluating the efficiency of our system, we compared it with some of the state-of-art algorithms Histogram-based [40], Dominant DCT [31], SIFT-SVD-Zernike [27], Fuzzybased [9] and DWT-CSLBP Based [23]. Table 3 Multiple combination of manipulations with its corresponding parameters To make the algorithms comparable, the images from Section 5.2 are being used. ...
Article
Full-text available
Perceptual image hashing methods utilize the visual phenomenon of the images and produce a fixed-length hash function and this hash value can be utilized for digital signature of an image. It can be used to show robustness against the digital manipulations done on the image and hence can be of use in different applications, viz., image indexing, tamper detection, etc. But to generate an efficient hash function is scarce as there is an inverse relationship of perceptual robustness and discrimination capability criteria. In this paper, we propose a robust and discrimination capable hash function by considering KAZE point feature descriptor for combinatorial manipulations. The KAZE detectors are used to find the stable key points of the image and then the three strongest regions are considered based on the strongest three key points of the image. Using these points, the features are generated and finally, the local features are used to generate the hash function and this hash function not only provides a good discrimination capable value with good robustness but also shows good results for double attacks and multiple combinations of attacks. Moreover, it outperforms the state-of-the-art algorithms in consideration for performances between discrimination capability and perceptual robustness.
... To design an image hashing algorithm, both robustness and discrimination are the main factors. There is a trade-off between these two properties, which means that if robustness is increased then discrimination is reduced and vice versa [9,10]. ...
... Gharde et al. [9] introduced a dual perceptual hash function to generate a hash using a fuzzy colour histogram. The L * a * b * colour space was used in this scheme. ...
... Let B 1 , B 2 , · · · , B n be the blocks of image and M 1 , M 2 , · · · , M n be the mean of each block. Let B 1 be the first block of the query image and a 1,1 , a 1,2 , · · · , a 16,16 be the raw pixel values of B 1 as shown in Equation (9). ...
Article
Full-text available
This work is focused on the development of a smart and automatic inspection system for printed labels. This is a challenging problem to solve since the collected labels are typically subjected to a variety of geometric and non-geometric distortions. Even though these distortions do not affect the content of a label, they have a substantial impact on the pixel value of the label image. Second, the faulty area may be extremely small as compared to the overall size of the labelling system. A further necessity is the ability to locate and isolate faults. To overcome this issue, a robust image hashing approach for the detection of erroneous labels has been developed. Image hashing techniques are generally used in image authentication, social event detection and image copy detection. Most of the image hashing methods are computationally extensive and also misjudge the images processed through the geometric transformation. In this paper, we present a novel idea to detect the faults in labels by incorporating image hashing along with the traditional computer vision algorithms to reduce the processing time. It is possible to apply Speeded Up Robust Features (SURF) to acquire alignment parameters so that the scheme is resistant to geometric and other distortions. The statistical mean is employed to generate the hash value. Even though this feature is quite simple, it has been found to be extremely effective in terms of computing complexity and the precision with which faults are detected, as proven by the experimental findings. Experimental results show that the proposed technique achieved an accuracy of 90.12%.
... Hence, tampering detection, a scheme that identifies the integrity and authenticity of the digital multimedia data, has emerged as an important research topic. Perceptual image hashing [1][2][3][4] supports image content authentication by representing the semantic content in a compact signature, which should be sensitive to content altering modifications but robust against content preserving manipulations such as blur, noise and illumination correction [5][6][7]. ...
Article
Full-text available
Image hashing-based authentication methods have been widely studied with continuous advancements owing to the speed and memory efficiency. However, reference hash generation and threshold setting, which are used for similarity measures between original images and corresponding distorted version, are important but less considered by most of existing models. In this paper, we propose an image hashing method based on multi-attack reference generation and adaptive thresholding for image authentication. We propose to build the prior information set based on the help of multiple virtual prior attacks, and present a multi-attack reference generation method based on hashing clusters. The perceptual hashing algorithm was applied to the reference/queried image to obtain the hashing codes for authentication. Furthermore, we introduce the concept of adaptive thresholding to account for variations in hashing distance. Extensive experiments on benchmark datasets have validated the effectiveness of our proposed method.
... Content based image authentication was first addressed by Schneider et al. [13] which uses image histogram to represent image content but the generated hash is very long and time consuming and also content of image can be changed without changing the histogram. Nilesh et al. [4] used a fuzzy color histogram to generate hash but it fails identify attacks related to pixel color value manipulation like contrast adjustment, brightness adjustment etc. Sobel, Canny [3] uses edges to represent content, the generated hash is long and cannot recover lost data. Storck et al. [15], Wu et al. [19,20], Lin et al. [10], Sun et al. [16], uses transform co-efficient like DCT, DWT etc., Sun et al. [16] uses wavelet transform to represent the image content and the generated hash cannot recover lost data. ...
Article
Full-text available
The image authentication plays a vital role in modern multimedia technology. The existing technique for preserving authentication abided the content only by preserving transformations like scaling, additive noise, gamma correction, brightness adjustments, water marking etc., but it fails to authenticate larger angles rotations. Very few systems had authenticated rotations to a smaller angle which was less than 5o. In existing system, the failure occurred in authenticating large angle rotation is due to the finite divisions of equal sized square block for calculating the local features. In this paper concept of dividing circular blocks with equal area is analyzed for better competency. The Haralick features are mostly calculated for square block. In the proposed system 14 features of Haralick has been grouped to hash code for each circular blocks in sender side. In receiver side, the same procedure is followed to generate hash code and the comparison is carried out to verify the authentication. In addition to scaling, brightness, contrast adjustment, gamma correction etc., the proposed system tolerates rotation even to greater angles up to 360o with better efficiency.
... Recently, the approaches authenticating images using their principle content, known as image hashing [15][16][17][18][19][20][21][22][23][24], emerge as the popular authentication techniques in video surveillance applications. Using image hashing approaches, the invariant features of the original image on the sender side are extracted and then represented as a numeric value called a hash. ...
Article
Due to large volume and high variability of editing tools, protecting multimedia contents and ensuring their privacy and authenticity has become an increasingly important issue in cyber-physical security of industrial environments, especially industrial surveillance. The approaches authenticating images using their principle content emerge as popular authentication techniques in industrial video surveillance applications. But maintaining a good trade-off between perceptual robustness and discriminations is the key research challenge in image hashing approaches. In this paper, a robust image hashing method is proposed for efficient authentication of keyframes extracted from surveillance video data. A novel feature extraction strategy is employed in the proposed image hashing approach for authentication by extracting two important features: the positions of rich and non-zero low edge blocks and the dominant DCT coefficients of the corresponding rich edge blocks, keeping the computational cost at minimum. Extensive experiments conducted from different perspectives suggest that the proposed approach provides a trustworthy and secure way of multimedia data transmission over surveillance networks. Further, the results vindicate the suitability of our proposal for real-time authentication and embedded security in smart industrial applications compared to state-of-the-art methods.
... A robust image hashing function produces similar hash values for images with same visual appearance but also is sensitive to content-changing distortions and malicious attacks. Robust image hashing received extensive attention in recent decades [3], [4]. Many image hashing algorithms are proposed and widely used in many fields including image authentication, digital watermarking, image retrieval, image copy detection, image quality assessment, and multimedia forensics [5]- [10]. ...
Article
Full-text available
This paper presents a robust image hashing algorithm that exploits low-rank decomposition and path integral local binary pattern (pi-LBP), referred to LRPL hashing. The proposed algorithm generates a compact binary sequence from a low-rank component of the normalized image as the hash code. Considering the excellent texture structure description ability of pi-LBP features, the new hashing algorithm extracts a feature vector from a low-rank feature matrix with pi-LBP. The pi-LBP feature vector is then encrypted by using the logistic map to produce the final hash sequence. The Hamming distance between the hash sequences is employed to authenticate the tested image. The experimental results demonstrate that the proposed LRPL hashing method has better robustness and discrimination compared with existing hashing algorithms.
Article
Content identification systems are an essential technology for many applications. These systems identify query multimedia items using a database of known identities. A hash-based system uses a perceptual hashing function that generates a hash value invariant against a set of expected manipulations in an image, later compared to perform identification. Usually, this set of manipulations is well-known, and the researcher creates the perceptual hashing function that best adapts to the set. However, a new manipulation may break the hashing function, requiring to create a new one, which may be costly and time-consuming. Therefore, we propose to let the hashing function learn an invariant feature space automatically. For this, we exploit the recent advances in self-supervised learning, where a model uses unlabeled data to generate a feature representation by solving a metric learning-based pretext task that enforces the robust image hashing properties for content identification systems. To achieve model transferability on unseen data, our pretext task enforces the feature vector invariance against the manipulation set, and through random sampling on the unlabeled training set, we present the model a wide variety of perceptual information to work on. As exhaustive experimentation shows, this method achieves excellent robustness against a comprehensive set of manipulations, even difficult ones such as horizontal flip and rotation, with excellent identification performance. Also, the trained model is highly discriminative against the presence of near-duplicate images. Furthermore, this method does not need re-training or fine-tuning on a new dataset to achieve the observed performance, indicating an excellent generalization capacity.
Chapter
We propose a novel perceptual hashing based on salient region and a nonnegative matrix factorization (NMF) in this paper. Firstly, the input image is standardized and processed with low-pass filtering. Secondly, the salient region is extracted from the obtained preprocessed image. The minimum bounding rectangle of each salient region is extracted, and the pixels in all rectangles are rearranged to form a secondary image, then the preprocessed image is decomposed by NMF to obtain the coefficient matrix as the final image hash. The algorithm is robust to general content-preserving manipulations through the experiment. The proposed algorithm outflanks some best in the performances of perceptual robustness and discrimination indicated by identification accuracy performances.KeywordsImage hashingSalient regionNMF
Article
Full-text available
In the recent digitization era, image hashing is a key technology, including image recognition, authentication and manipulation detection, among many multimedia security applications. The primary challenge in hashing schemes is to extract its robust feature. For a better understanding and design of a robust image hashing algorithm, it is very crucial to look into few important parameters like discrimination, robustness, reliability, etc. This paper reflects a detailed study of the existing literature on hashing-based image authentication techniques. This work provides a systematic overview and highlights the merits and demerits associated with various state-of-the-art techniques. In particular, the basic features and categories of image authentication techniques based on hashing are explored along with their properties. Moreover, different performance measures such as output metrices, receiver operating characteristics (ROC) parameters, execution time, etc., have been discussed in this work. The paper also compares the performances of various existing algorithms related to different content preserving operations on diverse data sets. This paper summarizes all the techniques and provides the most optimum solutions in regard to image hashing techniques based on different parameters.
Article
Image authentication is the technology of verifying image origin, integrity and authenticity. A rich stream of research on image authentication has shown various trade-off among four favorable features, namely robustness, security, versatility and efficiency. Image data authentication has the highest level of security but provides no robustness/versatility. Image content authentication from robust hashing keeps robust to limited types of operations, and as a result its versatility is not satisfactory. Existing designs of image content authentication from advanced cryptographic primitives achieve robustness, security and versatility, at the cost of low efficiency. In this paper, we present a new design of image authentication with an improved trade-off among the aforementioned features. Our versatile design is robust to a number of predefined image processing operations. Its security can be reduced to q-strong Diffie-Hellman (q-SDH), a complexity problem used by existing cryptographic algorithms. From the aspect of efficiency, the new design has a constant-size authentication overhead (⩽ 2 kB) and a constant verification time (around 0.05 s). While the time of generating authentication overhead increases linearly with the number of permissible editing operations, it only takes around 0.33 s for 1000 types of permissible operations. We believe the new design will facilitate image applications where trustworthy image editing is required.