Face Alignment in Thermal Infrared Images Using Cascaded Shape Regression

Area of Science:

• Thermography and Medical Imaging
• Computer Vision and Machine Learning

Background:

• Evaluating physiological and psychological states often relies on skin temperature from thermal infrared images.
• Accurate face alignment in thermal images is crucial for extracting relevant skin temperature data from regions of interest.
• Current methods like Active Appearance Model (AAM) for thermal face alignment are computationally expensive and lack real-time performance.

Purpose of the Study:

• To investigate the application of Cascaded Shape Regression (CSR) for face alignment in thermal infrared images.
• To evaluate the speed and robustness of CSR in thermal face alignment compared to existing methods.

Main Methods:

• Implementation and testing of Cascaded Shape Regression (CSR) for face alignment specifically on thermal infrared images.
• Comparative analysis of CSR performance against the Active Appearance Model (AAM) in terms of speed and robustness.

Main Results:

• Face alignment using CSR demonstrated superior robustness in thermal infrared images.
• CSR achieved significantly faster computation speeds compared to AAM for thermal face alignment.
• This indicates CSR's potential for real-time applications in thermal imaging analysis.

Conclusions:

• Cascaded Shape Regression (CSR) is a more efficient and reliable technique for face alignment in thermal infrared imaging.
• The findings suggest CSR can overcome the limitations of AAM, enabling practical real-time thermal imaging analysis for physiological assessments.