how can i use matlab for scene recognition with bag of words

Please i need some assistance in my project about scene recognition with bag of words, i want to use matlab for this project of mine but i don't know how to add the 100 images, how to extract features in all the images, how to apply kmeans to the feature centers, how to compute histogram of the images and finally to compare the histogram. I tried my possible best to do it but i still couldn't and time is against me, need an assistance from any one please, its my final project and have to be submitted latest on this coming Wednesday. my email address is [e-mail]. Thank you

Answer 1

[Image Analyst]

Here's an example of kmeans with images.

To sum images, cast to double first.

How you extract the features (segment the image) depends on what your images look like and what the features are that you want to segment. For example, if you want red spots then that will take a different feature detection algorithm than if you want gray areas with honeycomb patterns on them. There is no universal "extract features" algorithm that works with all images in the universe, in case that's what you were thinking (which is what it sounds like).

I'm not sure what the difficulty in using histograms is - share your code and tell us where you're confused. In the meantime, to process a hundred images, see the FAQ: http://matlab.wikia.com/wiki/FAQ#How_can_I_process_a_sequence_of_files.3F

Answer 2

[Dima Lisin]

There is now support for image classification using the bag of words model in the Computer Vision System Toolbox. See this example.