In this research, the image retrieval system is developed to search for images in the collection of images found on the database based on input (query) in the form of sample images provided by the user. The image is represented as a set of region segmentation results. Users of the system decides the object to be searched on the image and label an example of the region to train the system through a graphical interface. For the image retrieval process that contains a specific object, the user opens the image and choose the region containing the object on the system, then the system will rank all the regions in the collection are indexed query class to the collection in the database in the Euclidean distance (L2) of the region in user query and displays the results of the corresponding images from the collection on the database, which is the image that has the highest similarity.
Copyrights © 2014