Maha Charfeddine

A dynamic video watermarking algorithm in fast motion areas in the wavelet domain

In this paper, we propose a video watermarking algorithm which embeds different parts of a single watermark into different shots of a video under the wavelet domain. Based on a Motion Activity Analysis, different regions of the original video are separated into perceptually distinct categories according to motion information and region complexity. Thus, the localizations of the watermark are adjusted adaptively in accordance with the human visual system and signal characteristics, which makes them perceptually invisible and less vulnerable to automated removal. In addition, contrary to traditional methods where the watermark remains at a fixed position on the screen, the watermark moves along with moving objects and thus motion artefacts can be avoid. The multi-frame based extraction strategy ensures that the watermark can be correctly recovered from a very short segment of video. Individual frames extracted from the video also contain watermark information. Experimental results show that the inserted watermark is not only less perceptible but also robust against common video processing attacks.

A new DCT audio watermarking scheme based on preliminary MP3 study

In this paper, a new audio watermarking scheme operating in the frequency domain and based on neural network architecture is described. The watermark is hidden into the middle frequency band after performing a Discrete Cosine transform (DCT). Embedding and extraction of the watermark are based on the use of a back-propagation neural network (BPNN) architecture. In addition, the selection of frequencies and the block hiding the watermark are based on a preliminary study of the effect of MP3 compression at several rates on the signal. Experimental results show that the proposed technique presents good robustness and perceptual quality results. We also investigate the application of the proposed technique in video watermarking. Traditional techniques have used audio channel as supplementary embedding space and adopt state-of-the art techniques that resist to MP3 compression attack. In these techniques, the MPEG compression attack is only evaluated on the video part and the audio part is kept unaffected. In this paper, we adapt the preliminary MP3 study to video watermarking technique but with a preliminary study of the MPEG compression applied to the audio channel. Here again, we notice that the application of the preliminary MPEG study to the audio channel improves the robustness of the video watermarking scheme though keeping high-quality watermarked video sequences.

The impact of error correcting coding in audio watermarking

One of the most important issues of the digital watermarking is the watermarks robustness. Thats why the error correcting codes (ECC) techniques were proposed. In this article, the performance of ECC in audio watermarking system is researched. The conclusion is that BCH encoders and turbo codes are the most important encoders. They have the best experimental robustness results against several audio signal attacks.

A multimedia tracing traitors scheme using multi-level hierarchical structure for Tardos fingerprint based audio watermarking

This paper presents a novel approach in tracing traitors field. It proposes a multi-level hierarchical structure to the used probabilistic fingerprinting code; the well known Tardos code. This proposed structure is performed to address the problem of computational costs and time of Tardos code during its accusation step. The generated fingerprint is embedded in the extracted audio stream of the media by an audio watermarking technique operating in the frequency domain. The watermarking technique represents an original choice compared to the existing works in the literature. We assume that the strategy of collusion is known, we compare then the performance of our tracing traitors framework against different types of attacks. We show in this paper how the proposed hierarchy and the watermarking layer have a satisfying impact on the performance of our tracing system.

Clustering impact on group-based traitor tracing schemes

According to the ever development of multimedia distribution systems, more than one technique was proposed in the literature to address the copyright protection issue. One key technique was to propose a fingerprinting system based on traitor tracing process to retrieve back the traitorous users who can operate in the mid-way. Some previous works agree upon the fact that users belonging to the same group have more probability to collude together. Several researchers in the tracing traitor field agree upon the fact that constructing a group-based fingerprint should enhance the detection rates of the fingerprinting system. In this paper, we propose to generate a fingerprint having the group property by using a clustering algorithm. We propose to construct a group-based fingerprint according to a DCT-based audio watermarking technique which has proven good robustness and inaudibility results. To show the impact of the classifying algorithm, a set of experimental tests are conducted to check two relevant criteria: the capacity and the security of the group-based fingerprint.

A QR-code based audio watermarking technique for tracing traitors

Handling a great number of users and surviving different types of attacks present fundamental challenges of the majority fingerprinting systems in the tracing traitor field. In this paper, the proposed technique consists in embedding a fingerprint, a QR code in the audio stream extracted from the media release. Using the QR-code provides several advantages as supporting a large amount of information in a compact format end damage resiliency. This paper proposes to encode the identifier which is a parallel concatenation of two tracing codes: Boneh Shaw and Tardos codes into QR-code. The proposed approach should not only improve the two-stage tracing strategy by reducing the complexity computation, but also enhance the secure side of the proposed technique by the preprocessing treatment before generating the QR-code.

Towards a Blind MAP-Based Traitor Tracing Scheme for Hierarchical Fingerprints

One of the key challenges of a traitor tracing scheme is to deal with the real scenarios. In this context, the tracing operation is usually constrained by the lack of information about the number of colluders and even the collusion channel. Indeed, the Tardos decoding is invariant regardless the type of collusion, which can be considered as a suboptimality of its accusation performance. In this paper, we propose to use a MAP-based estimation strategy which improves the Tardos decoding step and guarantees a good estimation results. Compared to the original version of Tardos code and the original MAP decoder operating in blind tracing scheme, the proposed technique takes the advantage of operating in hierarchical context to provide a more concise and accurate accusation decision, this in a short time.

Audiovisual video characterization using audio watermarking scheme

Due to the incessant explosion of the multimedia documents amount, the use of metadata is becoming crucial to facilitate the retrieval and the management of these audiovisual contents. Metadata creation is highly time and resources consuming even if the process is automatically done. Thus, video browsing systems uses existing metadata files generally jointed to the corresponding video for efficient semantic multimedia content retrieval. However, missing the metadata file renders the related video useless. So, in this paper, a new strategy for video characterization is described by embedding the metadata information using a blind watermarking technique. Consequently, browsing systems can use the beforehand indexed content just by extracting the corresponding metadata.

Speaker Emotion Recognition: From Classical Classifiers To Deep Neural Networks

Speaker emotion recognition is considered among the most challenging tasks in recent years. In fact, automatic systems for security, medicine or education can be improved when considering the speech affective state. In this paper, a twofold approach for speech emotion classification is proposed. At the first side, a relevant set of features is adopted, and then at the second one, numerous supervised training techniques, involving classic methods as well as deep learning, are experimented. Experimental results indicate that deep architecture can improve classification performance on two affective databases, the Berlin Dataset of Emotional Speech and the SAVEE Dataset Surrey Audio-Visual Expressed Emotion.

Speaker gender identification based on majority vote classifiers

Speaker gender identification is considered among the most important tools in several multimedia applications namely in automatic speech recognition, interactive voice response systems and audio browsing systems. Gender identification systems performance is closely linked to the selected feature set and the employed classification model. Typical techniques are based on selecting the best performing classification method or searching optimum tuning of one classifier parameters through experimentation. In this paper, we consider a relevant and rich set of features involving pitch, MFCCs as well as other temporal and frequency-domain descriptors. Five classification models including decision tree, discriminant analysis, nave Bayes, support vector machine and k-nearest neighbor was experimented. The three best perming classifiers among the five ones will contribute by majority voting between their scores. Experimentations were performed on three different datasets spoken in three languages: English, German and Arabic in order to validate language independency of the proposed scheme. Results confirm that the presented system has reached a satisfying accuracy rate and promising classification performance thanks to the discriminating abilities and diversity of the used features combined with mid-level statistics.