A'bstract -Video is an important and challenge media and requires sophisticated indexing schemes for efficient retrieval from visual databases. An important step in video indexing is scene change detection. Recently, several scene change detection algorithms in the pixel and compressed(MPEG-2) domains have been reported in the literature. These algorithms are computationally complex and are not very robust in detecting gradual scene changes.In this paper, we propose an efficient technique for detecting scene changes i n1 the MPEG-2 compressed domain.The proposed algorithm has the advantage of fast scene change detection.In addition, this algorithm has the potential to detect gradual scene changes.

Diverse applications are used on mobile devices. Because of the increasing dependence on information systems, immense amounts of personal and sensitive data are stored on mobile devices. Thus, security or privacy breaches are a major challenge. To protect mobile systems and the private information on these systems from being accessed by adversaries, a framework for mobile user identification through the use of a multimodal behavioural biometrics scheme with a keystroke trajectory feature is presented herein. Conventionally, mobile devices have been protected by mechanisms such as PINs or passwords. However, these approaches have numerous disadvantages. Therefore, approaches that employ keystroke biometrics for secure and reliable mobile device identification have been proposed. Because unimodal behavioural biometrics identification mechanisms have limited accuracy and effectiveness, a multimodal scheme that includes different behavioural biometric traits, such as keystroke and swipe biometric traits, is examined. However, the information provided by the spatial and temporal features of keystroke biometrics is not comprehensive. Therefore, a trajectory model is derived to describe the behavioural biometric uniqueness of a user. In the user identification phase, a multistream recurrent neural network (RNN) is adopted. The results reveal that the proposed trajectory model performs well, and the multimodal scheme using an RNN with a late fusion method provides accurate identification results. The proposed system achieved an accuracy of 95.29%, F1 score of 94.64%, and equal error rate of 1.78%. Thus, the proposed mobile identification system is capable of resisting attacks that standard mechanisms may be vulnerable to and represents a valuable contribution to cyber security.

It is well known that orthogonal wavelet transform with filters of nonlinear phase gives poor visual results in low bit rate image coding. Biorthogonal wavelet is a good substitute, which is, however essentially nonorthogonal. A greedy steepest descend algorithm is proposed to design an adaptive quantization scheme based on the actual statistics of the input image. Since the L2 norm of the quantization error is not preserved through the nonorthogonal transform, a quantization error estimation formula considering the characteristic value of the reconstruction filters is derived to incorporate the adaptive quantization scheme. Computer simulation results demonstrate significant SNR gains over standard coding technique, and comparable visual improvements. L IntroductionRecently wavelet based image coding has become the domain of extensive research. For image coding application, it is desirable that an exact reconstruction subband coding scheme should correspond to an orthonormal basis with reasonable smooth mother wavelet. However, the resulting analysis and synthesis filters are in general not linear phase. On the other hand it is also desirable the FIR filter used be linear phase, since such filter can be easily cascaded in pyramidal filter structure without phase compensation [2]. Unfortunately, there are no nontrivial orthonormal linear phase FIR filters with the exact reconstruction property [5]. Relaxing the orthonormality requirement and using biorthogonal basis is a good alternative.As is well known, the L2 norm of quantization errors is not preserved through the nonorthogonal transform, so there does not exist an exact analytical formula for optimal bit allocation as in the case of orthogonal wavelet transform [3]. The goal of the paper is to improve the applicability of biorthogonal wavelet transform to image coding by proposing an adaptive quantization design scheme based on the statistics of the input image. The design procedure is based on a criterion which best describes the trade-off between distortion and rate, and hence results in superior performance. II. Biorthogonal Wavelets Siganl Analysis with Biorthogonal WaveletBiorthogonal wavelet bases were firstly introduced by Cohen [5] and were extensively used by numerous authors in image coding application. The basic idea of biorthogonal wavelet transform is to represent a signal f( x) as a superposition of two biorthogonal bases v,,,(x) = 2-m/2v(2-mx -n ) and $V',,,,(x) = 2-m'2v*(2-m'2x -n), where v*(x) is the dual basis of v(x), m and n are scaling and translation parameters respectively. Therefore the signal f ( x) can be expressed as and ic m , n ( f > = ( v m , n , f ) = I v , , n ( x ) f ( x W -_( 2 ) As in the orthonormal case, one can introduce multiresolution analysis. We then define two dual scaling functions: $,,, ( X) = 2-"12 $( 2-" x -n ) and its dual * 4 m . n

It is well known that there exists strong energy correlation between various subbands of a real-world image. A new powerful technique of Finite State Vector Quantization (FSVQ) has been introduced to fully exploit the self-similarity of the image in wavelet domain across different scales. Lattices in R have considerable structure, and hence, Lattice VQ offers the promise of design simplicity and reduced complexity encoding. The combination of FSVQ and LVQ gives rise to the so-called FSLVQ, which is proved to be succe:;sfill in exploiting the energy correlation across scales and simple enough in implementation.

No abstract