Santanu Halder

A fast FPGA based architecture for sobel edge detection

This paper presents an efficient FPGA based architecture for Sobel edge detection algorithm in respect of both time and space complexity. Various edge detection algorithms are typically used in image processing, artificial intelligence etc. In this paper the Sobel edge detection algorithm using hardware description language and its implementation in Field Programmable Gate Array (FPGA) device is presented in an efficient way. Sobel edge detection algorithm is chosen due to its property of less deterioration in high levels of noise. The result shows a significant improvement of time and space complexity over an existing architecture.

A proposed system for colorization of a gray scale facial image using patch matching technique

This paper proposes a novel approach to colorize a gray scale facial image with the selected reference face using patch matching technique. The colorization methodology has been applied on facial images due to their extensive use in various important fields like archaeology, entertainment, law enforcement etc. The experiment has been conducted with 150 male and female facial images collected from different face databases and the result has been found satisfactory. The proposed methodology has been implemented using Matlab 7.

Pattern Classification of Fabric Defects Using a Probabilistic Neural Network and Its Hardware Implementation using the Field Programmable Gate Array System

This study proposes a fabric defect classification system using a Probabilistic Neural Network (PNN) and its hardware implementation using a Field Programmable Gate Arrays (FPGA) based system. The PNN classifier achieves an accuracy of 98 ± 2% for the test data set, whereas the FPGA based hardware system of the PNN classifier realises about 94±2% testing accuracy. The FPGA system operates as fast as 50.777 MHz, corresponding to a clock period of 19.694 ns.

A novel unit circle approach for computation of sine function

This article proposes a novel method for Sine value computation and aims to design a fast FPGA based architecture for it. Unit Circle method for sine value calculation is a new one where a triangle is presumably located in the first quadrant of the circle of unit radius. Using the area calculation concept, the y coordinate of the third vertex of the triangle is approximated and this approximated value is the computed sine value. An FPGA-Based architecture for this novel unit circle approach is implemented in VHDL, synthesized on Xilinx Spartan 3 xc3s200-5ft256 FPGA kit simulated on ModelSim 6.2c. The proposed architecture gives accuracy of the computed sine value up to 15 bit in all cases in ten clock cycles. The maximum operating frequency of the proposed architecture is as fast as of 170.179 MHz.

Design of an Automatic Cartoon Movie Builder (ACMB) System for Generation of a Cartoon Movie from a Given Story

In this paper, a novel Automatic Cartoon Movie Builder (ACMB) system has been proposed to generate a cartoon movie based on a given story collected from the user and this concept is new in the literature. It has useful applications in cinematography and especially in the world of animation. The system has two types of databases: (a) Movie Database (DB-M) consisting of cartoon movies (animations) collected from different sources (b) Frame Database (DB-FR) consisting of shots (or sequence of frames depicting a continuous action) extracted from the movies of DB-M along with their textual characterization. The proposed system automatically generates a new cartoon movie from a given story with the help of DB-FR. In this paper, we focus on the generation of new cartoon movie with the retrieved shots from DB-FR as per the textual descriptions of a shot given by the user in the form of a story. The proposed method has been tested on 1000 shots which are extracted from 10 movies stored in DB-M.

A proposed system for cotton yarn defects classification using probabilistic neural network

Cotton yarn defect such as neps which are highly entangled fibres causes a serious problem in the textile industry. In this study, two types of cotton yarn neps, viz. seed coat and fibrous neps are classified by means of probabilistic neural network (PNN) using the features extracted from the images of neps. A k-fold cross validation technique has been applied to assess the performance of the PNN classifier. The results shows that the neps classification accomplished by means of image recognition through PNN classifier agree eminently well. The robustness, speed of execution, proven accuracy coupled with simplicity in algorithm holds the PNN as a foremost classifier to recognize yarn defects. The five fold cross validation is applied to measure the performance of the proposed method and it achieves nearly 96%-99% accuracy for the test data set.