Mahesh Goyani

Performance Analysis of Lip Synchronization Using LPC, MFCC and PLP Speech Parameters

Many multimedia applications and entertainment industry products like games, cartoons and film dubbing require speech driven face animation and audio-video synchronization. Only Automatic Speech Recognition system (ASR) does not give good results in noisy environment. Audio Visual Speech Recognition system plays vital role in such harsh environment as it uses both – audio and visual – information. In this paper, we have proposed a novel approach with enhanced performance over traditional methods that have been reported so far. Our algorithm works on the bases of acoustic and visual parameters to achieve better results. We have tested our system for English language using LPC, MFCC and PLP parameters of the speech. Lip parameters like lip width, lip height etc are extracted from the video and these both acoustic and visual parameters are used to train systems like Artificial Neural Network (ANN), Vector Quantization (VQ), Dynamic Time Warping (DTW), Support Vector Machine (SVM). We have employed neural network in our research work with LPC, MFCC and PLP parameters. Results show that our system is giving very good response against tested vowels.

Key Frame Detection Based Semantic Event Detection and Classification Using Heirarchical Approach for Cricket Sport Video Indexing

In this paper, we propose a key frame detection based approach towards semantic event detection and classification in cricket videos. The proposed scheme performs a top-down event detection and classification using hierarchical tree. At level 1, we extract key frames for indexing based upon the Hue Histogram difference. At level 2, we detect logo transitions and classify the frames as realtime or replay fragments. At level 3, we classify the realtime frames as field view, pitch view or non field view based on colour features such as soil colour for pitch view and grass colour for field view. At level 4, we detect close up and crowd frames based upon edge detection. At level 5a, we classify the close up frames into player of team A, player of team B and umpire based upon skin colour and corresponding jersey colour. At level 5b, we classify the crowd frames into spectators, player’s gathering of team A or player’s gathering of team B. Our classifiers show excellent results with correct detection and classification with reduced processing time.

Template matching and machine learning-based robust facial expression recognition system using multi-level Haar wavelet

ABSTRACT Recognition of facial expressions is important in industrial automation, security, medical, and many other fields. An image is a very rich and high dimensional data structure, which can result into a considerable computation when processed upon directly. Various feature extraction techniques have been proposed to represent the images efficiently in lower dimension which is understandable by the computer. In this paper, we propose Multi-Level Haar wavelet-based approach, which extracts salient features from prominent face regions at two different scales. The approach first segments most informative geometric components such as eye, mouth, etc. using the Adaboost cascade object detector. Segmented components are divided in M × N regions and feature vector is obtained by concatenating local Haar features extracted from each region. Feature vector is projected in Linear Discriminant Analysis space to reduce its size. For classification, we used template matching (Chi-Square and Cosine measure) and machine learning techniques (Logistic Regression and Support Vector Machine). Performance of proposed method is evaluated on various well-known data-sets like CK, Japanese Female Facial Expression, and Taiwanese Facial Expression Image Database. Adaptability of the feature is also tested on in-house Web-Enabled Spontaneous Facial Expression Data-set (WESFED). Comparison with state of the art method shows the superiority of proposed method.

Top Down Hierarchical Histogram Based Approach for Printed Devnagri Script Charecter Isolation

According to recent survey, there are at least 550 million people are using Devnagari script for communication. Hindi is one of the language, which is derived from Devnagari script. As it is a national language of India, Hindi Optical Character Recognition (OCR) System has a wide application in areas like post offices, Library automation, License Plate Recognition, Defense organization and many more government sectors. Research on printed and handwritten character recognition has been started before 50 years. Most of the research has been done for European texts. For any character recognition system, essential step is to identify individual character and find features to compare it with the template features. In this paper, we have proposed histogram based hierarchical approach for isolating individual character from the image document. For recognizing the characters, we can use Euclidean distanc, Pearson coefficient, chain code etc. Result shows that our system is quite robust and provides accuracy up to 92% for the charecter isolation.

Geometry and Skin Color Based Hybrid Approach for Face Tracking in Colour Environment

Face detection and face tracking has been a fascinating problem for image processing researchers during the last decade because of many important applications such as video face recognition at airports and security check-points, digital image archiving, etc. In this paper, we attempt to detect faces in a digital image using skin colour segmentation, morphological processing and acceptance/rejection based on face geometry. Face colour segmentation is done with different colour models like RGB and HSV for better results. The same algorithm is later on used for face tracking. Face tracking and video surveillance are some of the noticeable applications of the face detection. Face tracking in offline video is implemented using Skin Colour Segmentation algorithm. We have tested our system for standard face datasets CVL and LFW for face detection. Offline videos are recorded in current working environment with dynamic conditions. Results talk about the robustness of proposed algorithm.

Robust Face Recognition in Low Dimensional Subspace Using Reconstructive and Discriminative Features

In this paper, we have discussed dimensionality reduction techniques for face recognition - Principle Component Analysis (PCA) and Fisher Discriminant Analysis(FDA). Both the methods are based on linear projection, which projects the face from higher dimensional image space to lower dimensional feature space. PCA derives the most expressive features (MEF) by projecting face vector such that it captures greatest variance. FDA derives most discriminating features(MDF) by maximizing between class scatter and minimizing within class scatter. Lower dimensional features are used for recognition process. Classification can be achieved using Neural Network (NN), Support Vector Machine (SVM) etc. We have tested our system for the L2 norm measure. At the end of the paper, we have discussed results which show that FDA out weights the performance of PCA with average recognition rate more than 95%.