Anja Keskinarkaus

Audio watermarking using m-sequences and temporal masking

A novel algorithm for embedding a spread-spectrum based watermark into uncompressed, raw audio sequences is presented. The scheme takes advantage of temporal masking phenomenon in the human auditory system (HAS) in order to embed watermark data below the masking threshold of the audio signal. Detection of the watermark is done by blind detection, without using the original audio. As watermarking embedding and extraction are performed in the temporal domain, transformations to and from the frequency domain are not performed, neither on the embedding nor extraction part of the proposed scheme, resulting in the calculation simplicity of the embedding and detection process. The algorithm proved to be robust against mp3 compression (resulting in a BER of about 1.7%) and other common signal processing attacks, while the embedded watermark introduces no audible distortion of the host audio signal.

Multiple Domain Watermarking for Print-Scan and JPEG Resilient Data Hiding

In this paper, we propose a print-scan resilient watermarking method which takes advantage of multiple watermarking. The method presented here consists of three separate watermarks out of which two are used for inverting the geometrical transformations and the third is the multibit message watermark. A circular template watermark is embedded in magnitudes of the Fourier transform to invert rotation and scale after print-scan process and another template watermark is embedded in spatial domain to invert translations. The message watermark is embedded in wavelet domain and watermark robustness in both approximation coefficient and detail coefficients is tested. Blind, cross-correlation based methods are utilized to extract the watermarks. The obtained success ratios were at least 91% with JPEG and JPEG200 quality factors of 80-100 and scanner resolution of 300dpi. The BER obtained with the previous settings was less than 1.5%.

Toward an interactive poster using digital watermarking and a mobile phone camera

A method for embedding a watermark in print media, posters or other paper printouts and reading the watermark information blindly with a camera phone is proposed. A subtractive-additive embedding method is applied in which the message is coded with a directed periodic pattern. The message is detected and read by searching regularities in the autocorrelation function of a periodic signal. The robustness to disturbance occurring during printing process due to air interface and camera phone properties is ensured using noise reduction, modified JND model, enhanced peak detection with filtering and shaping and two-level coding of the message. The validity of the approach is proven with tests, and an application example of an interactive poster is examined.

Image watermarking with feature point based synchronization robust to print-scan attack

In this paper we propose a content based multibit watermarking method robust to print-scan attack. A method to extract feature points, robust in terms of watermarking, is proposed. The location of the watermark is tied to a coordinate system defined by robust feature points. A message sequence is mapped to a directional angle of periodic patterns, which are scattered and embedded into triangles in permuted locations. In watermark extraction, an interplay between feature extraction and watermarking ensures reliability and a multibit message can be decoded blindly from the locations pointed by the key. By detecting the alignment of the autocorrelations peaks and using a coding table, a multibit message can be extracted. Experiments show that the method provides robust and blind extraction of watermark information after a print-scan attack and a set of compound attacks.

Reading Watermarks from Printed Binary Images with a Camera Phone

In this paper, we propose a method for reading a watermark from a printed binary image with a camera phone. The watermark is a small binary image which is protected with (15, 11) Hamming error coding and embedded in the binary image by utilizing flippability scores of the pixels and block based relationships. The binary image is divided into blocks and fixed number of bits is embedded in each block. A frame is added around the image in order to overcome 3D distortions and lens distortions are corrected by calibrating the camera. The results obtained are encouraging and when the images were captured freehandedly by rotating the camera approximately -2 - 2 degrees, the amount of fully recovered watermarks was 96.3%.

Detection of short-term activity avalanches in human brain default mode network with ultrafast MR encephalography

Recent studies pinpoint visually cued networks of avalanches with MEG/EEG data. Co-activation pattern (CAP) analysis can be used to detect single brain volume activity profiles and hemodynamic fingerprints of neuronal avalanches as sudden high signal activity peaks in classical fMRI data. In this study, we aimed to detect dynamic patterns of brain activity spreads with the use of ultrafast MR encephalography (MREG). MREG achieves 10 Hz whole brain sampling, allowing the estimation of spatial spread of an avalanche, even with the inherent hemodynamic delay of the BOLD signal. We developed a novel computational method to separate avalanche type fast activity spreads from motion artifacts, vasomotor fluctuations, and cardio-respiratory noise in human brain default mode network (DMN). Reproducible and classical DMN sources were identified using spatial ICA prior to advanced noise removal in order to assure that ICA converges to reproducible networks. Brain activity peaks were identified from parts of the DMN, and normalized MREG data around each peak were extracted individually to show dynamic avalanche type spreads as video clips within the DMN. Individual activity spread video clips of specific parts of the DMN were then averaged over the group of subjects. The experiments show that the high BOLD values around the peaks are mostly spreading along the spatial pattern of the particular DMN segment detected with ICA. With also the spread size and lifetime resembling the expected power law distributions, this indicates that the detected peaks are parts of activity avalanches, starting from (or crossing) the DMN. Furthermore, the split, one-sided sub-networks of the DMN show different spread directions within the same DMN framework. The results open possibilities to follow up brain activity avalanches in the hope to understand more about the system wide properties of diseases related to DMN dysfunction.

Image watermarking with a directed periodic pattern to embed multibit messages resilient to print-scan and compound attacks

In this paper, we propose a blind method to embed multibit watermarks in images, robust to various attacks on geometry and print-scan process. For carrying the multibit information, a method using directed watermark patterns is proposed. A message sequence is mapped to a directional angle of a periodic pattern, which is then embedded into image blocks. In the detection, using autocorrelation function, filtering, masking, and adaptive line search with Hough transform, the alignment of the autocorrelations peaks are detected from image blocks, and interpreted as a message. The method provides a robust and blind extraction of information after a print-scan attack. Additionally, the experiments with two laser printer and two printout material show that the method is robust to compound attacks, such as Stirmark random bending, shearing, aspect ratio change, cropping, translation, or JPEG combined with print-scan.

MORE --- a multimodal observation and analysis system for social interaction research

The MORE system is designed for observation and machine-aided analysis of social interaction in real life situations, such as classroom teaching scenarios and business meetings. The system utilizes a multichannel approach to collect data whereby multiple streams of data in a number of different modalities are obtained from each situation. Typically the system collects a 360-degree video and audio feed from multiple microphones set up in the space. The system includes an advanced server backend component that is capable of performing video processing, feature extraction and archiving operations on behalf of the user. The feature extraction services form a key part of the system and rely on advanced signal analysis techniques, such as speech processing, motion activity detection and facial expression recognition in order to speed up the analysis of large data sets. The provided web interface weaves the multiple streams of information together, utilizes the extracted features as metadata on the audio and video data and lets the user dive into analyzing the recorded events. The objective of the system is to facilitate easy navigation of multimodal data and enable the analysis of the recorded situations for the purposes of, for example, behavioral studies, teacher training and business development. A further unique feature of the system is its low setup overhead and high portability as the lightest MORE setup only requires a laptop computer and the selected set of sensors on site.

Watermark recovery from a dual layer hologram with a digital camera

In this paper we present a method for reading a watermark from a dual layer hologram image with a digital camera. Here the base of the hologram out of which the hologram was designed is a binary image. The hologram obtained is placed on a planar surface and an image is captured from the hologram by setting up a light source and a digital camera on appropriate distances and angles relative to the hologram. The captured image is corrected from affine distortions and a saliency detector based segmentation is performed for the image. The watermark is read from the resulting reconstructed binary image. The obtained results show that the watermark can be recovered perfectly from a dual layer hologram with a proper setting of the camera and light sources.

Wavelet domain print-scan and JPEG resilient data hiding method

In this paper we present a print-scan resilient method to embed multibit messages into images. Multilevel watermarking principles are applied in order to embed a reference watermark and the message bits. Methods to embed a robust spatial domain reference watermark, utilizing HVS are proposed and methods to improve the estimation of affine transformation parameters from a periodic reference watermark are considered. The multibit message is embedded on the approximation coefficients of the wavelet transform, utilizing JND profile estimation and additive spread spectrum techniques. A blind correlation based detection method is made use of to recover the message. In the experiments the imperceptuality issues related to multilevel watermarking are considered with different parameter settings and the robustness against print-scan attack and JPEG compression is measured and the results are shown.