Volume 3 Number 4 March 2014
1 |
Adaptive Multiple Kernels with Sir-Particle Filter Based Multi Human Tracking For Occluded Environment Abstract: This paper proposes a new technique to build a fully automatic tracking system which handles occlusion problem in a complex environment. In multiple human tracking, handling of occlusion is the challenging issue. When occlusion occurs, kernel based tracking was proven to be the promising approach. Hence, to overcome the occlusion problem the human body was considered to have multiple kernels. In this paper, SIR-Particle filter tracking was embedded with multiple kernels that build a fully automatic tracking system. The accuracy of the tracking system was evaluated by using Multiple Object Tracking Accuracy (MOTA) metric. Our tracking system was experimented using PETS benchmark dataset and found that the accuracy was computed as 97%. |
2 |
Comparative Analysis of Skin Segmentation Methods in Image Mining Abstract: This paper presents the compression of color medical images with different color spaces. Even though multimedia data storage and communication technologies have attained rapid growth, compression of color medical images remains a challenging task. In the proposed method, color medical images are converted to different color spaces such as YCbCr, NTSC and HSV. Then decomposition of different color space image is done using curvelet transform. The decomposed images are then compressed using huffman coding. The results obtained for different color spaces were compared in terms of compression ratio and bits per pixel. |
3 |
A Novel Video Coding Technique for Robust Video Transmission Abstract: Now-a-days efficient video compression techniques are essential in order to make digital video applications feasible. A new approach to the video coding techniques is introduced here. For the efficient video transmission the Set Partitioning In Hierarchical Trees (SPIHT) algorithm is used. The three dimensional (3-D) SPIHT coder has proved its efficiency and its real-time capability in compression of video. Since Three Dimensional Spatio-temporal orientation trees coupled with powerful SPIHT sorting and refinement, it provides comparable performance to H.263 standard videos. The 3D wavelet transform (WT) algorithm is based on the “Group Of Frames” (GOF) concept. The group of frames are decomposed both temporally and spatially. The decomposition process utilizes the wavelet filters. The transform coefficients are coded using “Three Dimensional Set Partitioning in Hierarchical Trees” (3-D SPIHT). In the reconstruction phase, the 3-D SPIHT decoding algorithm and the inverse wavelet transform are employed respectively. |
4 |
Optical Disc Detection on Retina Image Using Genetic Hough Transform VLSI Optimization Technique Abstract: Hough transform is used for detecting circles in an image. To reduce the huge computations in Hough transform, a resource efficient architecture is essential. Resource efficient and reduction in processing time are achieved with data parallelism. We present a circle detection method based on genetic algorithms. Our genetic algorithm uses the encoding of three edge points as the chromosome of candidate circles (x, y, and r) in the edge image of the scene. Fitness function evaluates if these candidate circles are really present in the edge image. Our encoding scheme reduces the search space by avoiding trying unfeasible individuals, this result in a fast circle detector. The implementation of GA-based Hough transform on an FPGA. This architecture is implemented using alter a device at operating frequency of 200MHz. It compute the Hough transform of 512x512 test images with 180 orientation in 2.05 to 3.15ms with minimum number of FPGA resources. |
5 |
Efficient Test Pattern Generator for BIST using Multiple Single Input Change Vectors Abstract: Digital circuit’s complexity and density are increasing while, at the same time, more quality and reliability are required. These trends, together with high test cost, make the validation of VLSI circuits more and more difficult. This paper proposes a novel test pattern generator (TPG) for built-in self-test. Our method generates multiple single input change (MSIC) vectors in a pattern, i.e., each vector applied to a scan chain is an SIC vector. According to the different scenarios of scan length, this paper develops two kinds of SIC generators to generate Johnson vectors and Johnson codeword’s, i.e., the reconfigurable Johnson counter and the scalable SIC counter. The proposed TPG is flexible to both the test-per-clock and the test-per-scan schemes. A theory is also developed to represent and analyze the sequences and to extract a class of MSIC sequences. Analysis results have the favorable features of minimum transition of sequence, uniform distribution of pattern, uniqueness of pattern, and low hardware overhead. New seed generator circuit has to be developed to improve the fault coverage. |
6 |
Design of Multi bit flip flop in FIR application using clustering algorithm Abstract: In Integrated Circuit industry power has become a major contribution . The main attribute is the clock power in circuits of VLSI. In today’s VLSI design scenario, power utilization by clocking takes up a vital role especially in design that uses deeply scaled CMOS technology. Proficient power utilization tends to be an important constraint in modern IC design. The underneath idea of multi bit flip flop is to reduce the inverter number by sharing among flip flop. Indulging multi bit flip flop in synchronous design is becoming a considerable method for reducing clock power. The single bit flip flop cells uses a mutual number of inverter that possess high driving capability to drive over clock signal. Grouping of such cells to form multi bi flip flop can spare drive strength, dynamic power and area of common inverter where there is no compromise among the necessary constraint among area and power. In this paper, a Hausdorff clustering algorithm is utilized to obtain nearest clustering for merging flip flops. The multi bit technique is introduced in FIR circuit to lessen power as well as area. This satisfies with the above given constraints. According to the experimental results, our algorithm significantly reduces clock power by 25.8% and it is found that total gate count is reduced from 186 to 128. The delay is curtailed upto 1.19 ns which increases the speed. |
7 |
Power reduction Analysis using Dual threshold voltage Domino Logic Abstract: Dual threshold voltages domino design methodology utilizes low threshold voltages for all transistors that can switch during the evaluate mode and utilizes high threshold voltages for all transistors that can switch during the precharge modes. Employed standby switch can strongly turn off all of the high threshold voltage transistors which enhance the effectiveness of a dual threshold voltage CMOS technology to reduce the sub threshold leakage current. Sub threshold leakage currents are especially important in burst mode type integrated circuits where the majority of the time for system is in an idle mode. The standby switch allowed a domino system enters and leaves a low leakage standby mode within a single clock cycle. In addition, we combined domino dynamic circuits style with pass transistor XNOR and CMOS NAND gates to realize logic 1 output during its precharge phase, but not affects circuits operation in its evaluation and standby phase. The first stage NAND gates output logic 1 can guarantee the second stage computation its correct logic function when system is in a cascaded operation mode. The simulation results demonstrated with the help of MICROWIND software. carry look-ahead adder is designed at the transistor level with reduced chip area, power consumption and propagation delay time more than 40%, 45% and around 20%, respectively. |
8 |
Performance Evaluation of Partition and Hierarchical Clustering Algorithms for Protein Sequences Abstract: Bioinformatics is the use of computer technology for managing biological data and solving complex biological problems. Mining biological data provides the useful patterns from large datasets gathered in biology and in other related life sciences areas. Clustering of biological sequences into groups or families is necessary in genomics and proteomics. A significant number of algorithms and methods are available for clustering protein sequences. In this paper, we compare and evaluate the performance of two clustering algorithms namely K-means from partitioning method and agglomerative from hierarchical method for protein sequences. First, we describe each clustering methods and compare them through the validity indices and execution time as well. |
9 |
An Efficient Count Based Transaction Reduction Approach for Discovering Frequent Patterns Abstract: Apriori algorithm is a classical algorithm of association rule mining and widely used for generating frequent item sets. This classical algorithm is inefficient due to so many scans of database. And if the database is large, it will take too much time to scan the database. To overcome these limitations, researchers have made a lot of improvements to the Apriori. This paper analyses the classical algorithm as well as some disadvantages of the improved Apriori and also proposed two new transaction reduction techniques for mining frequent patterns in large databases. In this approach, the whole database is scanned only once and the data is compressed in the form of a Bit Array Matrix. The frequent patterns are then mined directly from this Matrix. It also adopts a new count-based transaction reduction and support count method for candidates. Appropriate operations are designed and performed on matrices to achieve efficiency. All the algorithms are executed in 5% to 25% support level and the results are compared. Efficiency is proved through performance analysis. |
10 |
Naïve Bayes Classification For Predicting Diseases In Haemoglobin Protein Sequences Abstract: The development of sequencing techniques led to an exponential growth of protein sequences in the public databases. The sequential information has been successfully applied to unveil the structures, functions, evolutionary relationships, etc. Lot of computational methods have been developed to classify the protein sequences and to predict the diseases based on their sequence information. The classification of biological sequences is one of the significant challenges in bioinformatics as well in genomics and proteomics. The existence of these sequence data in huge masses and their indistinctness and especially the high costs for lab experiments make use of data mining in disease prediction methods which are applied instead of laboratory experiments. Since a wide number of diseases are based on proteins and their sequences, the protein sequence analysis has been of great attention recently. The use of data mining techniques in protein sequence analysis provides an efficient way for examining the proteins to identify their characteristics and it also provides a way for better drug designing. In this research work, the hemoglobin protein based diseases are predicted by applying Naïve Bayes classifier. The performance of this classifier is analyzed by the factors classification accuracy and execution time. |