A: Data Collection.
1- Dataset collection: Collect hundreds of grayscale images (200 hundred is the minimum). The dataset should include a least 4 categories. (e.g cars, horses, buildings, tigers…).
The categories should be balanced (approximately the same number of images for each category).
2- Dataset labeling:
a. Assign a Label to each image. (E.g. For car image assign 1, horse 2, building 3 and tiger 4…).
b. After ordering your images, construct a vector containing the label of each image in same order of the images.
c. Save all images in a directory called ImagesDatayourNamed. Save the list of the image names ordred and its corresponding labeling vector in [url removed, login to view] and [url removed, login to view] files.
3- If needed, enhance or/and restore the images. Explain why it is need and the choice of the performed method.
B: Feature Extraction:
1- For each image from the image Data, extract at least one color feature.
2- Normalize you data. For each Feature, make each entry between 0 and 1 for all images.
3- Save each Feature in an MxN matrix where M is the number of images and N is the dimension of your Feature vector. You have to use the same order of the images as the one in your name and label files
4- Write a Readme File that describes each Feature used, indicates the corresponding name of the Feature file, and the name of the Matlab function that extract it.
1. Query Image Q (Selection and display): Select and display the query image.
2. Retrieved image Rank 1, 2,3 and 4: Display the resulting retrieved image.
3. Dist(Q,R1), Dist(Q,R2), Dist(Q,R3), Dist(Q,R4): display the distance between the query and the retrieved image under the retrieved one.
4. Score of this retrieval: evaluates the current [url removed, login to view] Report:-The final report should roughly have the following format:
Introduction - Motivation
Problem definition Proposed method
Experimentso Details of the experiments; observations
- you need to change all the images that to collect to grayscale before you start working.