فصلنامه پردازش علائم و داده ها
سال هفتم شماره 2 (پیاپی 14، پاییز و زمستان 1389)

This paper proposed an authentication system based on low quality palmprint. For implementation of this system, first the features is extracted by using Contourlet and wavelet transforms. In the second phase, some of the features are selected by using Across Group Variance (AGV) filter. In the last phase by using a classification method, the authentication is completed. For classification we evaluated three different methods, Support Vector Machine (SVM), Revised Nearest Neighbor (RNN), and Boosted Direct Linear Discriminant Analysis (BDLD). The experiment is performed on the famous PolyU Palmprint database. The results shows that by combination of the proposed system and BDLD classifier has better performance in comparison to other methods and the same database.

Periodic noises are repetitive patterns on digital images and decreased the visual quality of images. The various methods for reducing the effects of the periodic noise in digital images are firstly investigated. Then an intelligent median filter in the frequency domain with an acceptable computational cost is proposed. In the proposed method, the regions of noise frequencies are determined by analyzing intelligently the spectral of noisy image. Then only for the destroyed frequencies by periodic noise, a median filter with proper size in the frequency domain is applied and the spectrum of the image with reduced periodic noise is computed. The compared methods including the proposed method, the mean and the median filtering techniques, all in frequency domain are implemented not only under MATLAB environment, but also by C programming under OpenCV library. The results in different conditions show that the proposed filter shows higher performances, visually and statistically, and also in needs very lower computational cost.

Conventional word spotting systems determine hypothesized keywords and their confidence score using a speech recognizer. Acceptance or rejection of these keywords is intended based on comparison of their scores with a specific threshold. It has been proved that confidence score prepared by recognizer is highly dependent on sub-word structure of each keyword. So comparing assigned scores to keywords without considering their sub-word units could causes degradation in overall performance. In this paper a novel method for confidence score normalization is proposed which is based on sub-word units of each keyword and linear programming algorithm. In proposed method, a keyword-dependent correction term is added to the score of the keyword to maximize separation of confidence score histograms of true and false occurrences. Our results show a 2% improvement in FOM compared to baseline system. Also, choosing an appropriate feature vector has been discussed in this paper.

Concatenation of prerecorded speech segments is one of the speech synthesis methods, which is being named as unit selection based speech synthesis. The basic idea of such systems is to select the best speech units from a large database and concatenating them to produce the final waveform. This process usually is done based on minimizing a predefined cost function. Searching the best data sequence on a vast amount of data is a time-consuming process. Meanwhile, most of the applications cannot stand a long time for production of desired speech. In this paper, the beam search method is introduced to limit the search space and to increase the synthesis speed. In this method, the search tree is pruned to remove the paths which are less probable to be the optimum path. By so doing, the quality of synthesized speech is maintained and the time of synthesis is reduced. Prior to the results of experiments, removing 60% of search paths by beam search causes the time of synthesis to be reduced by 57%, decreasing the naturalness of speech from 4.2 to 4, based on MOS evaluation method. This decrease in the naturalness is equal to only 4.7%. Also by pruning the data based on the two approaches proposed in this paper, synthesis speed is increased by 10% and 14.07% respectively. The naturalness, however, is reduced by 5% and 3.8%.

Since money exchange is important in our daily life, many types of equipments such as Vending Machines, Currency Sorters, Automatic Teller Machines (ATM) and Currency Recognition systems for blind people have been made. More advanced devices with more capabilities are being made each day. As a result, efficient, fast and reliable currency recognition methods are required. Most currency recognition methods only use one attribute of currency images, such as major color, Ultra Violet spectrum or texture that are extracted from currency images. In this paper, we introduce a method for currency recognition that combines texture and color data together and applies them to a neural network. The best result from the other existing methods is at most 85 percent but our method shows 10 percent improvement compared to existing solutions.

This paper describes the design and evaluation of prosodically-sensitive concatenative units for a Persian text-to-speech (TTS) synthesis system. Thesyllables used are prosodically conditioned in the sense that a single conventional syllable is stored as different versions taken directly from the different prosodic domains of the prosodically labeled, read sentences. The three levels of the Persian prosodic hierarchy were observed in the syllable selection process, thereby selecting three different versions of each syllable from the prosodic domains of the intonational phrase (IP), accentual phrase (AP) and prosodic word (PW). A listening experiment designed to evaluate the quality of the syllable database showed that listeners preferred stimuli composed of prosodically appropriate diphones. We interpret this as supporting the view that segments carry prosodic domain information.