face recognition (different image size)
I have set of face images, but they have different size.I know I can resize all, but maybe there is scale invariant approach? And how can I align faces?
I have set of face images, but they have different size.I know I can resize all, but maybe there is scale invariant approach? And how can I align faces?
Try to use Local Binary Patterns Histograms, they don't require your input images to be of equal size:
And for face alignment, you can find a script in the chapter Aligning Face Images of the OpenCV Face Recognition tutorial at:
This should be easy to translate to any language you want, ssee Sorin's C# version for example:
For the face alignment problem you can detect the eyes using haar-cascade, calculate the angle and rotate the picture (Philipp Wagner has some excellent tutorials for this). I successfully used this method to align faces in my master thesis project (http://code.google.com/p/stereo-face-recognition/).
Asked: 2012-08-20 01:02:03 -0600
Seen: 1,994 times
Last updated: Aug 20 '12
How is Facerecognizer model stored?
Which confidence level and parameters for face recognizer?
How to categorize the images based on Illumination / shadow ?
model like ASM (active shape model) [closed]
How to reduce false positives for face detection
Is stereoRectifyUncalibrated efficient?