How to do the feature extraction on folders that contains a stack of images for codebook construction ?

I try to do features extraction on histological images, I would like to realize the feature extraction (MSER or SURF) on all the images in a folder to construct a codebook.

The main purpose would be to create a classifier (for example using k-means clustering) to check if the classifier is reliable.

I split my folder in training set (70%) and validation set (30%), I succed to use the different methods on 1 image, but I don't know how to do it on every images in order to create a codebook.

Thanks in advance,

Martin

[Walter Roberson]

http://matlab.wikia.com/wiki/FAQ#How_can_I_process_a_sequence_of_files.3F

[Christopher McCausland]

I would also have a look at datastores to make your life easier given the use case.

Answer 1

[Gojo]

Hey @Martin Briet,

I understand that you want to extract features from the images. You have succeeded with a single image and want to perform for all the images present in the directory.

You can refer to the following link as suggested by @Walter Roberson: https://matlab.fandom.com/wiki/FAQ#How_can_I_process_a_sequence_of_files.3F

You can also have a look at the case:

https://www.mathworks.com/matlabcentral/answers/62399-how-to-load-multiple-images-and-processing-them

You can use “fullfile” to get image names and loop through them using “imread” function. You can use functions like “detectMSERFeatures” or “detectSURFFeatures” to extract features and then store them in a matrix. Then you can perform kmeans clustering algorithm.

Hope this helps!!