Distinguished Professor, Department of Computer Science， Michigan State University
When: Friday, June 17, 2011. 9:00 AM.
Approximate duration of 1.5 hours
Where: Room 314, Tower B, Fundamental Teaching Building
What: Automatic face recognition has received increased attention over the past ten years due to its important role in addressing a variety of security concerns, namely, access control to secure areas, de-duplication of passports and driver licenses and identification of suspects in surveillance videos. While there has been a substantial improvement in the performance of face recognition systems, they still do not meet the expectations and requirements of many forensics, law enforcement and commercial applications. The difficulties in automatic face recognition arise from the large intra-class variability in the face images due to many intrinsic (expression, aging) and extrinsic variations (pose, ambient illumination). In other words, different face images of the same person acquired at different times and under different imaging conditions can have substantially different appearances that cannot be correctly matched by state of the art techniques. The challenge is to a design salient feature extractor and a robust matcher. This talk will address a number of ongoing research projects in my lab that illustrate the importance and challenges on face recognition. These include (i) matching face sketch to photo, (ii) age-invariant face recognition, (iii) face recognition at a distance, (iv) heterogeneous face recognition, and (iv) use of micro-level features (face marks).