DOC PREVIEW
UW-Madison ECE 738 - Face Sketch Recognition

This preview shows page 1-2 out of 6 pages.

Save
View full document
View full document
Premium Document
Do you want full access? Go Premium and unlock all 6 pages.
Access to all documents
Download any document
Ad free experience
View full document
Premium Document
Do you want full access? Go Premium and unlock all 6 pages.
Access to all documents
Download any document
Ad free experience
Premium Document
Do you want full access? Go Premium and unlock all 6 pages.
Access to all documents
Download any document
Ad free experience

Unformatted text preview:

Face Sketch RecognitionProblem StatementGeometric MethodsEigen-Sketch MethodPhoto-to-sketch transformationResultsECE738 Advanced Image ProcessingFace Sketch Recognition [Tang04] IEEE Trans. CSVT, Jan 2004(C) 2005 by Yu Hen Hu2ECE738 Advanced Image ProcessingProblem Statement•Goal:–Match face images to face sketches–Useful when test faces are in sketch forms (e.g. no photo of suspect).•Approach–Transform face images to sketches–Match sketches of faces using an eigen-sketch approach(C) 2005 by Yu Hen Hu3ECE738 Advanced Image ProcessingGeometric Methods•Fiducial grid model–Identify anchor points tip of noise, corners of mouth–Derive other fiducial points from the identified anchor point•Geometric features–35 fiducial points used–26 distances between key fiducial points measured.–Including sizes of nose, eyes, eyebrows, face contours, etc.•Measure geometric features from both face image and sketches and perform matching(C) 2005 by Yu Hen Hu4ECE738 Advanced Image ProcessingEigen-Sketch Method•Convert face image into eigenface representation.•Generate eigen-sketch from each eigenface.•Project probe sketches (test sketches) onto the subspace spanned by eigen-sketches. •Use feature vectors of eigen-face representation to match the weight vectors •Assumption: –Face images and sketches are normalized in size, lighting, poses, and expression neutral–No occlusions on test images (eye glasses, beard, etc.)(C) 2005 by Yu Hen Hu5ECE738 Advanced Image ProcessingPhoto-to-sketch transformationPhoto-to-sketch transformation examples. (a) Original photo. (b) Reconstructed photo. (c) Reconstructed sketch. (d) Original sketch.Photo – eigen-face – eigen-sketch - sketch(C) 2005 by Yu Hen Hu6ECE738 Advanced Image ProcessingResults•Using Feret test set, the proposed method (sketch transform) out-performs conventional method (geometric and eigen-face) by large margin. •Possible reason: –Geometric face: features measured on photo may be different from sketches–Eigenface: gray scale image value differ very much between phton and sketch.Rank 1 2 3 4 5 6 7 8 9 10Geometric 30 37 45 48 53 59 62 66 67 70Eigenface 31 43 48 55 61 63 65 65 67 67Sketch transform 71 78 81 84 88 90 94 94 95


View Full Document

UW-Madison ECE 738 - Face Sketch Recognition

Download Face Sketch Recognition
Our administrator received your request to download this document. We will send you the file to your email shortly.
Loading Unlocking...
Login

Join to view Face Sketch Recognition and access 3M+ class-specific study document.

or
We will never post anything without your permission.
Don't have an account?
Sign Up

Join to view Face Sketch Recognition 2 2 and access 3M+ class-specific study document.

or

By creating an account you agree to our Privacy Policy and Terms Of Use

Already a member?