this is for holding javascript data
Miguel Tuazon edited To_detect_the_face_the__.tex
about 8 years ago
Commit id: 1256600ccb4dcd9324bf0b12cf97296f03cf2001
deletions | additions
diff --git a/To_detect_the_face_the__.tex b/To_detect_the_face_the__.tex
new file mode 100644
index 0000000..e059f5d
--- /dev/null
+++ b/To_detect_the_face_the__.tex
...
The face area obtained from the detection algorithm applied in the RGB image, \citet{gonzalez2014kinect} extract the same area from the depth image for the face detection. Four preprocessing stages were applied before the facial feature detection. These preprocessing stages are: edge removal, depth increase, median filter, and low-pass filter. Though the face region in the depth image is fitted with an analytic surface (fitting information mainly to a paraboloid), it is confirmed that both regions are almost identical, with the exception of the edges where there are discontinuities due to this there are a great differences. Thus, the start of the preprocessing stage is the removal of the edges in the face area decreasing 30 percent of the area of the rectangle given by the face detection. Also to make irregularities of the surface smaller depicted in the depth image, a primary smoothing is applied through a median filter, for which the depth value of a pixel is the average of the neighboring pixels. as the last stage of the preprocessing, a low-pass filter is applied to smooth the image. A 3 x 3 convolution mask with weights equals to 1/9 was chosen. Following the HK classification, image pixels could be labeled as belonging to a viewpoint-independent surface class type based on the combination of the signs from the mean and Gaussian curvatures as demonstrated on table 3 \cite{gonzalez2014kinect}.
