Thermal Images corresponding to the facial images [20] Different studies conducted by several researchers concluded that a pattern of unique heat distribution can be produced from the human face. This pattern can be analyzed and studied by using infrared cameras. The heat pattern, also called as ―heat signature‖ of a human face is produced with bone densities, blood vessels and skin. The example of thermal images is the Equinox database [20] which is the collection of facial images in the different modalities. An example of Equinox image database is explored in Figure 1 [20]. Generally there are nine parameters, which are considered from the human face image for the construction of thermogram. Nose and ears are not considered as a parameter for the thermogram construction. When the face image is captured, its thermogram image is matched with the thermal graphs which are already stored corresponding to every image. The matching procedure is based on Monte Carlo analysis, which measures the performance. The Matching Performance is enhanced by considering the visible and thermal images together. In other words, the fusion of visible and thermal images increases the matching performance. In 2003, Socolinsky DA conducted a study in which a matching performance comparison was shown of the two face recognition algorithms i.e. eigenfaces and Arena on to the Long-Wave Infrared and visible face image set [21]. The procedure was applied on the infrared videos of 91 subjects, in which the maximum classification performance was achieved 99% for ARENA on Long Wave Infrared (LWIR) Imagery while the minimum score achieved was 97% while the matching performance of eigen faces on LWIR imagery, was 96% and 87% for the same training sets. In 2004 Buddharaju P, conducted a face recognition methodologies based on multiple appearance for which the input image was considered as the combination of visible and thermal images [22] and concluded that the thermal images produce the better result. The Recognition accuracy also increased when the same algorithm was used on to the combination of visible and thermal face images. The face recognition matching was performed by using Bayesian classifier on the Equinox database, resulting the higher matching rate 89.6%. 

Context in source publication

Context 1

... 1: Thermal Images corresponding to the facial images [20] Different studies conducted by several researchers concluded that a pattern of unique heat distribution can be produced from the human face. This pattern can be analyzed and studied by using infrared cameras. The heat pattern, also called as ―heat signature‖ of a human face is produced with bone densities, blood vessels and skin. The example of thermal images is the Equinox database [20] which is the collection of facial images in the different modalities. An example of Equinox image database is explored in Figure 1 [20]. Generally there are nine parameters, which are considered from the human face image for the construction of thermogram. Nose and ears are not considered as a parameter for the thermogram construction. When the face image is captured, its thermogram image is matched with the thermal graphs which are already stored corresponding to every image. The matching procedure is based on Monte Carlo analysis, which measures the performance. The Matching Performance is enhanced by considering the visible and thermal images together. In other words, the fusion of visible and thermal images increases the matching performance. In 2003, Socolinsky DA conducted a study in which a matching performance comparison was shown of the two face recognition algorithms i.e. eigenfaces and Arena on to the Long-Wave Infrared and visible face image set [21]. The procedure was applied on the infrared videos of 91 subjects, in which the maximum classification performance was achieved 99% for ARENA on Long Wave Infrared (LWIR) Imagery while the minimum score achieved was 97% while the matching performance of eigen faces on LWIR imagery, was 96% and 87% for the same training sets. In 2004 Buddharaju P, conducted a face recognition methodologies based on multiple appearance for which the input image was considered as the combination of visible and thermal images [22] and concluded that the thermal images produce the better result. The Recognition accuracy also increased when the same algorithm was used on to the combination of visible and thermal face images. The face recognition matching was performed by using Bayesian classifier on the Equinox database, resulting the higher matching rate 89.6%.  ...