1 / 12

Face Detection

Face Detection. --- EE368 Project Presentation Ajay Gupta , Gang Xie and Jiwu Tang Stanford University 30 May, 2002. Integration of Segmentation and Template Matching Approach. Skin-Color based Segmentation. Template Matching. Refine. Skin-Color Based Segmentation.

lok
Télécharger la présentation

Face Detection

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Face Detection --- EE368 Project Presentation Ajay Gupta, Gang Xie and Jiwu Tang Stanford University 30 May, 2002

  2. Integration of Segmentation and Template Matching Approach Skin-Color based Segmentation Template Matching Refine

  3. Skin-Color Based Segmentation • RGB to YCbCr conversion • Generate a binary image using the chrominace components of the skin color • Morphological Processing on the binary image to fill up the holes and remove small isolated regions • Remove some of the small regions having area less than a certain threshold value • Remove some of the non-face regions using certain feature values such as orientation and ratio of MajorAxisLength to MinorAxisLength.

  4. Segmentation – Original Image

  5. Segmentation – Image after Morphological Processing

  6. Segmentation – Image after First Level of Thresholdings

  7. Segmentation – Image after Second Level of Thresholdings

  8. Segmentation – Final Segmented Image

  9. Face Template Construction Step 1: Extract 169 faces from 7 training images based on the ground truth data Step 2: Decide the template size (20*20) based on statistics Step 3: Manually form a training set containing 26 faces Step 4: Manually measure the positions of the center of two eyes and the center of mouth for each face Step 5: Scale all faces to the same size Step 6: Move all faces to the same position Step 7: Apply intensity histogram equalization on each face while keep hue and saturation unchanged Step 8: Average, resize, subtract its mean, and flip

  10. Template Matching s, t size(s) > size(t) ? no yes conv2(s, t) peak > threshold ? no yes record peak return • Inputs: • image s(x,y) • template t(-x,-y) scale s down

  11. Template Matching Further Eliminates Skin-Colored Non-Faces After segmentation After template matching

  12. Testing Results Image False No. Hits Positive Score ---------------------------------------------- 1 23 0 23 2 20 0 20 3 24 0 24 4 18 0 18 5 25 0 25 6 24 1 23 7 19 2 17

More Related