NCADSIP 2015- National Conference On Advances In Digital Signal And Image Processing

"NCADSIP 2015 Conference Papers "

Efficient Blocking Artifacts Removal on Color image using DCT Domain[ ]


Image enhancement techniques involve processingan image to make it look better to human viewers. Now,increasingly images are being represented in the compresseddomain format for efficient storage and transmission. Henceit has become imperative to investigate compressed domaintechniques to eliminate the computational overheads occur inspatial domain techniques. Processing in the DCT domain hasattracted significant attention of researchers due to its adoptionin the JPEG and MPEG compression standards. The imagesneed to be enhanced in terms of brightness along with colorsrestored with minimal computational overhead. This techniquetreats chromatic components in addition to the processingof the luminance component also removes blocking artifactsefficiently for improving the visual quality of the images to agreat extent.

------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Score Level Fusion Based Personal Authentication Using Fingerprint and Speech[ ]


In this paper development of a multimodal based biometric fusion system is discussed. A fingerprint recognition system is developed using global singularity features. Mel-frequency Cepstral Coefficients are used to recognise a speaker using the backpropagation artificial neural network. A score level fusion based recognition system is developed using fingerprint and speech match scores and the equal error rate (EER) measured shows a good improvement with 100% recognition rate is obtained for over a large span of match score threshold.

------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Implementation of L2 Way Tagged Cache Architecture[ ]


The advancements of semiconductor technology have boosted the rapid growth of very large scale integrated (VLSI) systems in our day-to-day life. Microprocessors and systems-on-chip (SOCs) are now extensively used in a variety of applications ranging from smart phones to handheld computers, from entertainment systems to sophisticated automotive controllers, and from gaming devices to life-saving medical equipment. The processing speed or performance of these systems is primarily limited by the power budget, which is determined by the battery life for mobile devices. Many high-performance microprocessors employ cache write-through policy for performance improvement and at the same time achieving good tolerance to soft errors in on-chip caches. However, write-through policy also incurs large energy overhead due to the increased accesses to caches at the lower level (e.g., L2 caches) during write operations. In this paper, we propose a new cache architecture referred to as way-tagged cache to improve the energy efficiency of write-through caches. By maintaining the way tags of L2 cache in the L1 cache during read operations, the proposed technique enables L2 cache to work in an equivalent direct-mapping manner during write hits, which account for the majority of L2 cache accesses. This leads to significant energy reduction without performance degradation. Furthermore, the idea of way tagging can be applied to existing low-power cache design techniques to further improve energy efficiency.

------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Wavelet Energy based Satistical Learning Approaches to Vocoid Consonant Recognition[ ]


State – of – the – art Automatic Speech Recognition (ASR) employs rigorous experimental evaluations on large, standard corpora from the real world. In recent years ASR and Machine Learning (ML) algorithms have had a great deal of influences on each other and feature selections can be considered as an essential task in ML. Compared with traditional basic speech feature extraction techniques, Wavelet Transform (WT) are highly capable of interpreting information content of the signal. This paper focuses on the recognition of Malayalam Vocoid Consonant (VC) speech units, a unique characteristic of the Malayalam language, using WT based Wavelet Energy (WE) parameters to capture the acoustical properties of each speech units. In the classification stage ML based on statistical approaches using with k – Nearest Neighbor (k – NN) is implemented. From the experimental results it is reported that k-NN algorithm can be perform well with wavelet family db5 compared with others in speaker independent environment.

------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Single image super resolution using learned wavelets-Block wavelet method[ ]


This method is based on patch based approach. It has many advantages compared with many wavelet transform based methods where DWT is applied on whole image and used parent child relation to predict the high resolution image. Results show that the new method outperforms different existing methods in terms of SNR values and visual quality

------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
A FACE RECOGNITION SYSTEM BY HIDDEN MARKOV MODEL AND DISCRIMINING SET APPROACH[ ]


Different approaches have been proposed over the last few years for improving holistic methods for facerecognition. Some of them include color processing, different face representations and image processingtechniques to increase robustness against illumination changes. There has been also some research about thecombination of different recognition methods, both at the feature and score levels. Embedded hidden Markov model(E-HHM) has been widely used in pattern recognition.The performance of Face recognition by E-HMM heavily depends on the choice of model parameters.In this paper, we proposea discriminating set ofmulti E-HMMs based face recognition algorithm. Experimental results illustrate that compared with the conventional HMMbased face recognition algorithm the proposed method obtainbetter recognition accuracies and higher generalization ability.

------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------