Please use this identifier to cite or link to this item:
http://hdl.handle.net/10603/431792
Title: | Investigations on human emotion recognition via facial expressions using feature extraction techniques and deep learning models |
Researcher: | Saranya, R |
Guide(s): | Poongodi, C |
Keywords: | Computer Science Computer Science Information Systems Engineering and Technology Facial expression recognition Static image Statistical Shape Projection Model |
University: | Anna University |
Completed Date: | 2021 |
Abstract: | Emotions are important in human communication to develop a wellbeing newlinesociety. In past few years, researchers developed various facial newlineexpression recognition (FER) system to interpret emotions as the demand of newlinehuman computer application increases. But developing an FER system in close newlineto human reality is still challenging due to unconstrained imaging conditions. newlineThe research work develops an FER system based on machine newlinelearning and deep learning approaches for static image and video-sequences. newlineIn machine learning approach, Statistical Shape Projection Model (SSPM) and newlineEnhanced Multi-Feature Fusion Model (EMFM) are the two novel methods newlineproposed in relation to geometric and appearance based feature extraction newlinetechniques, respectively. In deep learning approach, Maximum Boosted newlineConvolutional Neural Network (MBCNN) and integrated model of MBCNN newlineand LSTM are proposed in this research work to achieve high expression newlineclassification accuracy in recognizing emotions. newlineStatistical Shape Projection Model (SSPM) obtains high newlineclassification accuracy by extracting features using statistical shape model and newlineIntegral Projection analysis. The static image is considered as an input and the newlineedge details are enhanced using unsharp masking. The statistical shape model newlinefits a new image to the model and the deformation are learnt quickly. The newlineproposed feature extraction technique performs better after it is combined with newlineprojection analysis. The prominent edge features are improved with an newlineenhancement technique which helps in finding the discriminant features newlineeffectively. newline |
Pagination: | xviii,145p. |
URI: | http://hdl.handle.net/10603/431792 |
Appears in Departments: | Faculty of Information and Communication Engineering |
Files in This Item:
File | Description | Size | Format | |
---|---|---|---|---|
01_title.pdf | Attached File | 10.49 kB | Adobe PDF | View/Open |
02_prelim pages.pdf | 1.11 MB | Adobe PDF | View/Open | |
03_content.pdf | 479 kB | Adobe PDF | View/Open | |
04_abstract.pdf | 8.31 kB | Adobe PDF | View/Open | |
05_chapter 1.pdf | 617.41 kB | Adobe PDF | View/Open | |
06_chapter 2.pdf | 542.13 kB | Adobe PDF | View/Open | |
07_chapter 3.pdf | 930.72 kB | Adobe PDF | View/Open | |
08_chapter 4.pdf | 617.11 kB | Adobe PDF | View/Open | |
09_chapter 5.pdf | 856.59 kB | Adobe PDF | View/Open | |
10_chapter 6.pdf | 717.75 kB | Adobe PDF | View/Open | |
11_annexures.pdf | 180.22 kB | Adobe PDF | View/Open | |
80_recommendation.pdf | 119.16 kB | Adobe PDF | View/Open |
Items in Shodhganga are licensed under Creative Commons Licence Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0).
Altmetric Badge: