Ersin Esen

Forbidden Zone Data Hiding

A new blind data hiding method is proposed based on a novel concept of forbidden zone, where no alteration is allowed in a host signal during message embedding step. Depending on the desired probability of error, the range of the forbidden zone varies, as a compromise between robustness and embedding distortion. Hence, the proposed method makes use of this zone through a single control parameter in conjunction with modulating quantizers for message embedding. The superiority of the proposed scheme over QIM is presented theoretically, as well empirically via simulations. The method is further compared with DC-QIM and its weaknesses and strengths are stated.

Data hiding using trellis coded quantization

Information theoretic tools lead to the design and analysis of new blind data hiding methods. A novel quantization-based blind method, which uses trellis coded quantization, is proposed. The redundancy in initial state selection during trellis coded quantization is exploited to hide information as the index of this initial state. This index is recovered at the receiver by Viterbi decoding after comparison with all initial states. The performance of the proposed method is compared against other well-known approaches via simulations and promising results are obtained. Based on these results, the proposed method can be preferred in certain applications with high distortion attacks.

Fight detection in surveillance videos

Fight detection is an important topic for surveillance systems. However, there has been little success in creating an algorithm that can detect fight in surveillance videos with high performance. In this work, we propose a new method for the task of fight detection in surveillance videos. The proposed method relies on a novel motion feature, namely Motion Co-Occurrence Feature (MCF). Firstly, motion vectors are extracted by using block matching algorithm. Secondly, direction and magnitude values of motion vectors are quantized separately. Afterwards, direction and magnitude based MCF is calculated by considering both current and past motion vectors. Experimental results obtained using k-Nearest Neighbor classifier showed that the proposed algorithm can discriminate fight scenes with significantly high accuracy.

Comparison of Quantization Index Modulation and Forbidden Zone Data Hiding for compressed domain video data hiding

Data hiding is now a part of daily life through various applications. In this work, we apply two data hiding methods, Quantization Index Modulation and Forbidden Zone Data Hiding, to video applications. We place these methods into a general video data hiding scheme and compare their performance against compression attacks. The results of the experiments with typical TV content indicate the superiority of FZDH, specifically for powerful attacks.

Gender Classification via Gradientfaces

In this paper illumination invariant, pose and facial expression tolerant gender classification method is proposed. A recently introduced feature extraction method, namely Gradientfaces, is utilized together with Support Vector Machine (SVM) as a classifier. Image regions obtained from cascaded Adaboost based face detector is used at the feature extraction step and faster classification is achieved by using only 20-by-20 pixel region during feature extraction. For performance evaluation, two well-known face databases, FERET and Yale B are tested and the algorithm is compared against a pixelbased algorithm on these datasets. The results indicate that Gradientfaces significantly outperform the pixel-based methods under severe illumination, pose and facial expression variances.

A MPEG-7 compliant Video Management System: BilVMS

In the recent years, there has been a growing interest towards the management of multimedia data. Recently, ISO MPEG organization has finished establishing a new standard, MPEG-7, for describing audio-visual data for interoperable indexing, searching and browsing purposes. Following this standard, a state-of-the-art video management system has been designed and implemented. The system is capable of temporally segmenting video into shots, as well as obtaining a semantically meaningful group of shots, i.e. scenes. The scene decomposition is achieved using a HMM-based formulation by multimodal features. Keyframes are used as shot representatives and their visual descriptions are utilized to make similarity queries. Moreover, these low-level descriptors are also used to reach a number of semantic visual classes using support vector machines. Finally, automatic detection of human faces via skin color filtering and videotext recognition increase the indexing capabilities of the BilVMS system.

Comparison of Forbidden Zone Data Hiding and Quantization Index Modulation

Developing practical data hiding schemes that approach theoretical steganographic capacity limits is still a challenging research area. In this aspect, this problem is approached from a different perspective by proposing a novel method that relies on the newly introduced forbidden zone concept, which is defined as the host signal range, where no alterations are allowed. The proposed method, Forbidden Zone Data Hiding (FZDH), combines the forbidden zone with binning schemes and it is formulated both in general and simple parametric forms. This novel scheme is compared against the canonical binning-based data hiding schemes, such as QIM and DC-QIM, empirically, as well as in a theoretical framework. The results indicate that FZDH outperforms QIM, whereas performing competitive with DC-QIM.

Effects of noises in images used for height map reconstruction of cartridge cases using photometric stereo

Photometric stereo (PS) is commonly used in examination of ballistic marks and this study investigated the effects of noises present in images on the reconstructed depth map based on PS and the sensitivity of different error measures to different kinds of noises. The images of 3D synthetic objects with known light conditions and Lambert reflection model were calculated first. Thermal noise, Schottky noise and speckle noise with different amplitudes were added to these images. A common effect seen in images, the vignetting effect, was also modelled. Surface normals were calculated using these noisy images and the height maps were reconstructed. Normalized absolute error and normalized gradient error were calculated for all cases and the results were compared.

Content based video copy detection with coarse features

Content Based Copy Detection is an alternative approach to invisible watermarking for tracking duplicate data. Primary stages are creating a database using the features belonging to the original data and searching query data in terms of its features in this database. Features must be robust against targeted attacks and discriminative enough to distinguish different content. In this work, we propose reducing the precision of feature values to attain robustness and increasing the number and dimension of features to attain discriminativity. To this end, we create a feature database using different features, which correspond to different information sources, together. We detect the original sources of the query videos in this database, which is composed of coarse features, by feature comparison. Effectiveness of the proposed method against various attacks is observed through experiments.

Voting system based robust and efficient audiocopy detection

Audio Copy Detection(ACD) problem has several difficulties due to signal distortions and huge amount of audio data to be searched. In this paper we propose a fast audio copy detection system which is very robust against common signal distortions.The proposed method performs a vote based search on 15 bit representation of audio data.