2.5D Facial Expression Recognition using Photometric Stereo and the Area Weighted Histogram of Shape Index

HEp-2 Cell Classification Using Shape Index Histograms With Donut-Shaped Spatial Pooling

“…In recent years it has gained popularity for range imaging where it is used as a robust surface descriptor, typically for visual recognition [17], [18]. In some cases, the shape index is summarized in histograms [19]- [21] following the trend of the popular local image features like SIFT and histograms of oriented gradients (HOG) [22]. A related approach to texture description is basic image features (BIF) [23] based on firstand second-order differential structure.…”

Robust 3D face capture using example-based photometric stereo

“…Capture of the three-dimensional geometry of the human face is 11 a computer vision problem with many applications, including 12 facial recognition and mood detection, computer animation, 13 plastic surgery and automated sculpture. In particular, it has been 14 shown in recent years that the use of the 3D geometry of the face 15 improves the robustness of facial recognition methods under 16 variations in illumination, pose and perspective [1][2][3][4].…”

“…As a potential 3D facial capture method, its main 40 advantages are that it requires very simple and inexpensive 41 equipment, can be used in ordinary environments without 42 hampering the subject's motion or demanding his cooperation, 43 and can capture high-resolution 3D data in a fraction of a second 44 [10][11][12]. Indeed, Broadbent et al demonstrated a photometric stereo 45 system that captures 640 Â 480 depth maps at video rates (15 46 frames-per-second) using a PC with a popular graphics card [13]. 47 Photometric stereo does not obtain the depth information 48 directly; instead it measures the average normal of the surface 49 within each image pixel.…”

“…These 51 integrated depths are somewhat less accurate than those obtained 52 with laser ranging and structured light, but are accurate enough for 53 facial recognition, considering the natural variation of human face 54 geometry over time. Indeed, the slope data can be used directly in 55 facial recognition algorithms [14,15]. 56 Photometric stereo requires information on the finish of the 57 surface and on the illumination conditions of each input image.…”