IJIGSP Vol. 7, No. 10, Sep. 2015
Cover page and Table of Contents: PDF (size: 664KB)
Conventional approaches to sampling images use Shannon theorem, which requires signals to be sampled at a rate twice the maximum frequency. This criterion leads to larger storage and bandwidth requirements. Compressive Sensing (CS) is a novel sampling technique that removes the bottleneck imposed by Shannon's theorem. This theory utilizes sparsity present in the images to recover it from fewer observations than the traditional methods. It joins the sampling and compression steps and enables to reconstruct with the only fewer number of observations. This property of compressive Sensing provides evident advantages over Nyquist-Shannon theorem. The image reconstruction algorithms with CS increase the efficiency of the overall algorithm in reconstructing the sparse signal. There are various algorithms available for recovery. These algorithms include convex minimization class, greedy pursuit algorithms. Numerous algorithms come under these classes of recovery techniques. This paper discusses the origin, purpose, scope and implementation of CS in image reconstruction. It also depicts various reconstruction algorithms and compares their complexity, PSNR and running time. It concludes with the discussion of the various versions of these reconstruction algorithms and future direction of CS-based image reconstruction algorithms.[...] Read more.
Contour representation of binary object is increasingly used in image processing and pattern recognition. Chain code and crack code are popular methods of contour encoding. However, by using these methods, an accurate estimate of geometric features like area and perimeter of objects are difficult to obtain. Mid-crack code, another contour encoding method, can help to obtain more accurate estimate of the geometric features of objects. Though a considerable amount of reduction of the size of images is obtained by fixed-length mid-crack code, yet, more efficient encoding is possible by considering and applying variable-length encoding technique. In this paper, a compressed mid-crack code is proposed based on the Huffman code. Experiments performed on different images yield that the proposed representation reduces the number of bits require to encode the contour of an image with compared to the classical mid-crack code.[...] Read more.
Recently, speech recognition (SR) has drawn a great attraction to the research community due to its importance in human-computer interaction bearing scopes in many important tasks. In a SR system, acoustic modelling (AM) is crucial one which contains statistical representation of every distinct sound that makes up the word. A number of prominent SR methods are available for English and Russian languages with Deep Belief Network (DBN) and other techniques with respect to other major languages such as Bangla. This paper investigates acoustic modeling of Bangla words using DBN combined with HMM for Bangla SR. In this study, Mel Frequency Cepstral Coefficients (MFCCs) is used to accurately represent the shape of the vocal tract that manifests itself in the envelope of the short time power spectrum. Then DBN is trained with these feature vectors to calculate each of the phoneme states. Later on enhanced gradient is used to slightly adjust the model parameters to make it more accurate. In addition, performance on training RBMs improved by using adaptive learning, weight decay and momentum factor. Total 840 utterances (20 utterances for each of 42 speakers) of the words are used in this study. The proposed method is shown satisfactory recognition accuracy and outperformed other prominent existing methods.[...] Read more.
In this research paper, we have developed a system that identifies users by their voices and helped them to retrieve the information using their voice queries. The system takes into account speaker identification as well as speech recognition i.e. two pattern recognition techniques in speech domain. The conglomeration of speaker identification task and speech recognition task provides multitude of facilities in comparison to isolated approach. The speaker identification task is achieved by using SVM where as speech recognition is based on HMM. We have used two different types of corpora for training the system. Gamma tone cepstral coefficients and mel frequency cepstral coefficients are extracted for speaker identification and speech recognition respectively. The accuracy of the system is measured from two perspective i.e. accuracy of speaker identity and accuracy of speech recognition task. The accuracy of the speaker identification is enhanced by adopting the speech recognition at the initial stage of speaker identification.[...] Read more.
Face recognition is one of the important and popular visual recognition problem due to its challenging nature and its diverse set of applications. That's why face recognition is attracted by many researchers. Methods based on Local Binary Pattern (LBP) are widely used for face recognition in the literature, and it is sensitive to noise. To address this present paper utilized the powerful local texture descriptor that is less sensitive to noise and more discriminant in uniform regions called as Local Ternary Pattern (LTP). The Uniform Local Binary Pattern (ULBP) derived on LBP treats a large set of LBP under one label called as miscellaneous. This may result some loss of information on LBP and LTP based methods. To address this two Prominent LBP (PLBP) are derived, namely PLBP-Low (L) and PLBP-High (H) on LTP. Based on this the present paper derived eight texture features on facial images. A distance function is used on proposed texture features for effective face recognition. To eliminate most of the effects of illumination changes that are present in human face an efficient preprocessing method is used that preserves the significant appearance details that are needed for face recognition. The present method is experimented on Yale, Indian and American Telephone and Telegraph Company (AT&T) Olivetti Research Laboratory (ORL) data bases and it has given state-of-the-art performance on the three popular datasets.[...] Read more.
Technological limitations restrict to acquire an image at high spatial and high temporal resolutions with space borne global sensors. In this paper, we propose a novel technique to create such images at ground-based data processing system. The Resourcesat-2 is one of the Indian Space Research Organization (ISRO) global missions and it carries Linear Imaging and Self-Scanning Sensors (LISS III and LISS IV) and an Advanced Wide-Field Sensor (AWiFS). The spatial resolution of LISS III is 23.5 m and that of AWiFS is 56 m. The temporal resolution of LISS III is 24 days and that of AWiFS is 5 days. Objective of the paper is to create a synthetic LISS III image at 23.5 m spatial and 5-day temporal resolutions. A synthetic LISS III image at time tk is created from an AWiFS image at time tk and a single AWiFS–LISS III image pair at time t0 which is acquired before or after the prediction time tk , here t0≠tk. The proposed method involves three phases. The first is super resolution phase. In this phase, two transition images are obtained for the time t0 and tk by improving AWiFS spatial resolution. The second is high pass modulation phase. In this phase, the high frequency details which are obtained in the difference of LISS III image and the transition image of time t0 are proportionally injected into the transition image at time tk. In composition of multi-temporal images of different spatial resolutions, spurious spatial discontinuities are inevitable. In the third phase, these spurious discontinuities are identified and smoothed with the spatial-profile-averaging method. The proposed method achieves better prediction accuracy when compared to the state-of-the art techniques.[...] Read more.
Biometric system is an alternative way to the traditional identity verification methods. This research article provides an overview of recently / currently used single and multiple biometrics based personal identification systems which are based on human physiological (such as fingerprint, hand geometry, head recognition, iris, retina, face recognition, DNA recognition, palm prints, heartbeat, finger veins, footprints and palates) and behavioral (such as body language, facial expression, signature verification and speech recognition) characteristics.
This paper focuses on RGB based palatal pattern analysis of persons and the proposed technique uses RGB values with silhouette computes of palatal patterns for identifying a person. We have tested our proposed technique for palatal patterns of 50 persons including males & females and it is observed that RGB values based silhouette technique are accurately identifying the persons on the basis of their palatal patterns. For each person seven palatal images were taken. Out of these seven palatal images, four images were used for training dataset and last three palatal patterns were used for identifying the persons. The proposed technique is reliable & secure and it is a foolproof method which is clearly differentiating the persons on the basis of their palatal patterns.
Performing digital image forgery is very easy due to highly precise image editing tools. There is a concomitant need to have some mechanism to differentiate between a forged image and the original image. In this paper, we propose a passive image forgery detection method that uses entropy filter and local phase quantization (LPQ) texture operator. The entropy filter generally highlights the boundary of the forged regions. It is due to the fact that the entropy filter provides the randomness of a pixel in its local neighborhood. The LPQ operator provides internal statistics of the image based on the phase information. We apply entropy filter on different sized neighborhoods followed by LPQ operator on the CASIA v1.0, CASIA v2.0 and Columbia image forgery evaluation databases. We consider these databases in our experiments because these are standard databases and have been used in most of the methods. Our method provides promising results on both CASIA databases; however, they are comparable on Columbia database with that of the existing state of the art methods.[...] Read more.