Algorithms for invariant long-wave infrared face segmentation: evaluation and comparison

This paper presents two methods for automatic segmentation of images of faces captured in long wavelength infrared, allowing a wide range of face rotations, expressions and artifacts (such as glasses and hats). We also present the validation of segmentation results using a recognition method to show the impact of the segmentation accuracy on the recognition. The paper presents two different approaches (one aimed at real-time performance and the other at high accuracy) and compares their performance against three other previously published methods. The proposed approaches are based on statistical modeling of pixel intensities and active contour application, although several other image processing operations are also performed. Experiments were performed on a total of 893 test images from four public available databases. The obtained results improve on previous existing methods up to 29.5 % for the first measure error (E 1) and up to 34.7 % for the second measure (E 2), depending on the method and database. Regarding the computational time, our proposals improve up to 63.32 % when compared with the other proposals. We also present the validation of the various segmentation methods that are presented by applying a face recognition method.