Thursday, June 23, 2005
Suggestions
Email for Dr Rueckert:
After spending my time doing nothing but reading, I have gathered a few
interesting facial expression recognition research areas. Based on my reading so far, I have noticed that many of these research works acquire 2D images to identify and classify the face expressions. Some use 2D or 2.5D images to model and synthesize facial expressions. Most of the face expression animation works are based on Facial Action Coding Systems (FACS) and Action Units (AUs). Well, there are other methods used too. Some research have incorporated both 2D and 3D face data but I have yet to find and read anything on 3D facial expression recognition. Nevertheless I will still continue searching and reading.
One of the research areas that I thought of looking into is recognising facial expression from 3D data based on holistic and local approach and then identify the effectiveness of the methods used.
Or I can keep to the initial proposal that is to deploy a deformable model. I am planning to look at using either ASM or AAM onto 3D model. The deformable model will be used to normalise the acquired 3D face data (with or without expressions) to a standard or neutral expression so that matching can be done easily.
Suggestions?
After spending my time doing nothing but reading, I have gathered a few
interesting facial expression recognition research areas. Based on my reading so far, I have noticed that many of these research works acquire 2D images to identify and classify the face expressions. Some use 2D or 2.5D images to model and synthesize facial expressions. Most of the face expression animation works are based on Facial Action Coding Systems (FACS) and Action Units (AUs). Well, there are other methods used too. Some research have incorporated both 2D and 3D face data but I have yet to find and read anything on 3D facial expression recognition. Nevertheless I will still continue searching and reading.
One of the research areas that I thought of looking into is recognising facial expression from 3D data based on holistic and local approach and then identify the effectiveness of the methods used.
Or I can keep to the initial proposal that is to deploy a deformable model. I am planning to look at using either ASM or AAM onto 3D model. The deformable model will be used to normalise the acquired 3D face data (with or without expressions) to a standard or neutral expression so that matching can be done easily.
Suggestions?
# posted by Jacey Lynn Minoi : 11:55 am
