3D Face Modeling from Multi-Spectral Imaging (Research Proposal Sample)

3D Face Modeling from Multi-Spectral Imaging
Names
Department
University
City, Country
Abstract
This research proposal intends to simulate a three dimension (3D) face modeling by merging multispectral imaging techniques with time of flight cameras (ToF) so as to come up with an effectual algorithmic scheme that will mold 3D meshes to automatically plan textures from images with the intent of coming up with an advanced recognition accuracy. This research equally exploits the diverse applications of such a model to forensic studies as well as medical field. A sensor system of three dimension multi-spectrum model combining diverse optical sensors and 3D depth sensor will be adequately presented in this paper. Diverse image sensors such as 3D and infrared (IR) sensors will be put into scrutiny. The accuracy of facial recognition has previously been altered by diverse facial expressions a subject may arrive at. Therefore, in order to diminish the above discrepancy; this paper intends to exhibit an extension to 3D face recognition algorithm. This will raise its efficiency for datasets from subjects comprising or not comprising a neutral facial appearance by up to 50 percent.
Introduction
In the current world, human face entails vital information, most natural as well as least invasive biometrics. It’s worth stipulating that numerous institutions such as banking sector and airports check-ins have taken chances with biometric-based identifications and authentication measures in ensuring efficient and safety in delivering their mandates. Therefore, the relevance of this research proposal comes into play by introducing such a model into realizing recognition accuracy. Studies have provided how facial expressions have hindered smooth performance of face recognition systems to such a level that some states currently prompts the subjects requesting for passports to pose with normal expression. This proposal intends to eliminate this hindrance courtesy of the proposed model which intends to attain extremely high verification rates despite of any present facial expressions.[S. HÄusken,et al, Strategies and benefits of fusion of 2d and 3d face recognition," technical report, Viisage Technology AG, 2005.]
Practically, images from different cameras are used to restructure the entire 3D face geometry of the individual (subject), unto which the data from the infrared camera is plotted to. Having reviewed previous literature on this area of study, this paper intends to come up with a brand similarity metric which raises the efficiency and toughness of our model with respect to facial expressions. This model intends to exhibit better results on datasets as compared to the existing models with respect to presence of facial expressions. However this remains valid as long as only 3D data is employed. The research paper will be categorized into several sections namely review of previous literature, methodology of our proposed model, results of the model, and finally the implications and applications.
Review of previous literature
* Face recognition
This paper intends to cover only the recent studies in this field. Milan et al illustrates coefficient of linear correlation by use of tensor depiction for 3D faces. Nevertheless, the shown results portray only 85% rank-1 recognition with respect to 277 subjects on the database. Another study by Buciu and Nafornita depicts on the disadvantages of using 2D images in recognition and identification of subjects, such as illumination changes. The proposed model intends to counter this hindrance by merging multispectral imaging techniques with time of flight cameras (ToF) so as to come up with an effectual algorithmic scheme that will mold 3D meshes to automatically plan textures from images with the intent of coming up with an advanced recognition accuracy. Another study by Tan and Briggs had a model of a local ternary pattern (LTP) which acted an improved model for the LBP. This model comprised of high sensitivity and illumination within the region of the captured image. This aided in normalizing facial and pose expressions that would otherwise vary on thermal-IR images.[Milan,A Bennamoun,M and Owens,R, Matching tensors for pose invariant automatic 3d face recogni-tion," in Proc. IEEE Workshop on Advanced 3D Imaging for Safety and Security, (San Diego, CA), 2005.] [Buciu, I.Nafornita, I Non-negative matrix factorization methods for face recognition underextreme lighting variations. In Proceedings of International Symposium on Signals, Circuits and Systems, (ISSCS 2009), Iasi, Rumania, 9–10 July 2009; pp. 1–4.] [Tan,X Triggs, B, Enhanced local texture feature sets for face recognition under difficult lighting conditions. Trans. Image Process. 2010, 19, 1635–1650.]
Methodology
This project proposal intends to use the following summarized criteria in coming up with a successful 3D Face Modeling...