How to improve accuracy rate on Eigenface Algorithm

I use EmguCV library(openCV wrapper) on C# to implemented face detection and facial recognition using Eigenface algorithm

I found many mistakes of accuracy rate such as minimum distance is not a same person in Training Set, person who doesn't exist in Training Set but match with person in Training set with good distance, etc.

please help me.

Thank you.

additional information: - I use 1 facial image per 1 p开发者_开发技巧erson in Training set (straight facial image, no orientation) - Now, I test with around 10-20 people in Training set

Try blurring the images. Since you are only using one training image per person, Eigenface does not have a way to tell whether the high-frequency component or low-frequency component would be more useful.

To decide on how much to blur / downscale the images, do an experiment with a human. Give a person some blurred test image pairs and ask them to match. Try to find out how much you can blur the images and still be successful at matching them. Use the same blurring parameters for the computer.

Image alignment techniques may also help. For example, normalize the images so that the tip of nose, and the two eyes (forming a triangle) are affine-transformed to a standard set of locations.

Finally, some simple eigenvalue preprocessing techniques or image preprocessing techniques (such as histogram equalization) may help.

• Below are the following options which may help you boost your accuracy

1] Image Normalization:

• Make your image pixel values from 0 to 1

2] Image Alignment (This is a very important step to achieve good performance):

• Align all the train images and test images so that eyes, nose, mouth of all the faces in all the images have almost the same co-ordinates

• Check this post on face alignment (Highly recommended) : https://www.pyimagesearch.com/2017/05/22/face-alignment-with-opencv-and-python/

3] Data augmentation trick:

• You can add filters to you faces that will have an effect of the same face in different lighting conditions
• So from one face you can make several images in different lighting conditions

4] Removing Noise:

• Before performing step 3 apply Gaussian blur to all the images