Arda Ustubioglu

Medical Image Tamper Detection Based on Passive Image Authentication

Telemedicine has gained popularity in recent years. Medical images can be transferred over the Internet to enable the telediagnosis between medical staffs and to make the patient’s history accessible to medical staff from anywhere. Therefore, integrity protection of the medical image is a serious concern due to the broadcast nature of the Internet. Some watermarking techniques are proposed to control the integrity of medical images. However, they require embedding of extra information (watermark) into image before transmission. It decreases visual quality of the medical image and can cause false diagnosis. The proposed method uses passive image authentication mechanism to detect the tampered regions on medical images. Structural texture information is obtained from the medical image by using local binary pattern rotation invariant (LBPROT) to make the keypoint extraction techniques more successful. Keypoints on the texture image are obtained with scale invariant feature transform (SIFT). Tampered regions are detected by the method by matching the keypoints. The method improves the keypoint-based passive image authentication mechanism (they do not detect tampering when the smooth region is used for covering an object) by using LBPROT before keypoint extraction because smooth regions also have texture information. Experimental results show that the method detects tampered regions on the medical images even if the forged image has undergone some attacks (Gaussian blurring/additive white Gaussian noise) or the forged regions are scaled/rotated before pasting.

DCT based image watermarking method with dynamic gain

Digital watermarking is a basic method for copyright protection. This paper presents a blind watermarking approach in DCT domain with Spread Spectrum technique. The method divides the image into non-overlapping blocks of size 8×8. Each block is transformed into frequency domain using DCT. Middle band frequency coefficients of each block are utilized to embed and extract watermark. Two number sequences of size 1×22 are carefully (first set in increasing order and the other one in decreasing order) predetermined by the algorithm to improve the robustness of the method against some attacks. The strength of the watermark is also determined dynamically according to the energy characteristic of the current block. Experimental results show that the proposed method can extract watermark even when an image was distorted by some attacks such as JPEG compression, cropping and noise addition. The method also has higher Normalized Cross-Correlation (NC) values compared to other works in the literature.

LBP-DCT Based Copy Move Forgery Detection Algorithm

Increase on the availability of the image editing software makes the forgery of the digital image easy. Researchers proposed methods to cope with image authentication in recent years. We proposed a passive image authentication technique to determine the copy move forgery. First, the method divides the image into overlapping blocks. It uses LBP (Local Binary Pattern) to label each block. Labeled blocks are transformed into frequency domain using DCT (Discrete Cosine Transform). Sign values of the first fifteen coefficients of the zigzag scanned block plus average Y, Cb, Cr values constitutes the feature vector for the block. Finally, the feature vectors are lexicographically sorted and element-by-element similarity measurement is used to determine the forged blocks. Experimental results show that the method has higher accuracy ratios and lower false negative values under some post processing operation compared to other DCT based methods. Our method can also detect multiple copy move forgery.

A new watermark algorithm resistance to geometric transformations

The increase in the number of image processing tools to edit digital images and their ease of use have driven the need for image authentication techniques. A technique called Digital Watermarking based on hiding special codes in image has gained popularity among all other authentication techniques. An active research topic is to reconstruct the watermark even after a watermarked image has been processed by known attacks. Key point extraction algorithm, SURF, is used to determine pixels to hide and recover the watermark. Quantization index modulation technique is used during watermark embedding into pixel intensity values and recovery. A watermarking method based on SURF has been proposed as an alternative to SIFT based methods in the literature. Proposed method generates considerably high normalized correlation against attacks as shown in the results.

Automated pre-diagnosis of Acromegaly disease using Local Binary Patterns and its variants

Symptoms of some disease are seen on the face area. These symptoms make typical faces by giving typical characteristics to face areas and expressions. Hippocratic face, Parkinson face, Lupus face, Leprosy face can be given as example of these faces. One of the most typical ones of these faces is Acromegaly face. Acromegaly is a disease which occurs as a result of secretion of excessive amounts of growth hormone (GH). In this work, we propose a new and effective system that can pre-diagnose of Acromegaly automatically by the way of evaluating the patients face images. For this purpose, Local Binary Patterns (LBP) and its modified models Improved Local Binary Patterns (ILBP), Center Symmetric Local Binary Patterns (CS-LBP) are applied for feature extraction of face images. Weighted Chi-square, Euclidean and Manhattan classifiers are used for the classifying to the selected sets of features. Our results showed that LBP (8,1) coupled with Manhattan classifiers resulted in highest accuracy of 97%, sensitivity of 93%, specificity of 100% compared to other feature extraction techniques and classifiers. In this way, our proposed system is more suitable for diagnosis of Acromegaly disease with higher accuracy.

A fast detection method for frame duplication forgery based on correlation

In recent years, fast development of video editing software has made video forgery applicable. One of the most frequently encountered forgery types in video forensics is the frame duplication forgery. Researches have proposed methods to deal with this type of forgery. The two main drawbacks of this methods reported in the literature are execution time and low detection accuracy. In this work a new frame duplication forgery detection method that uses correlation between neighboring frames to extract features from video is proposed. Experimental results show that the proposed method has lower execution time with better detection accuracy compared to similar works reported in the literature.

ACromegaly Pre-Diagnosis Based On Principal Component And Linear Discriminant Analysis

Acromegaly is a disease which occurs as a result of secretion of excessive amounts of growth hormone. Today, accurate diagnosis of Acromegaly can be diagnosed using the result of clinical and biochemical tests and measurements. Before applying these tests, the experts use the physical appearance of patience to diagnose. As a result of this, pre-diagnosis varies depending upon the experience of the doctors and patients are subjected to lots of necessary and unnecessary tests. In this study, using the patients face images, a new and efficient software has proposed which automatically diagnoses Acromegaly invariant of age, gender and facial expression. For this purpose, after applying various pre-processing to face images, Principal Component Analysis (PCA), Linear Discriminant Analysis (LDA) and Subspace Linear Discriminant Analysis (Subspace LDA) are used to extract features and Euclidean and Manhattan classifiers are used to classify the obtained features. Our results showed that TBA+DAA coupled with Euclidean resulted in highest accuracy of 96%, sensitivity of 100%, specificity of 95% compared to other feature extraction techniques.