A Learning State-Space Model for Image Retrieval

A Learning State-Space Model for Image Retrieval

Blog Article

This paper proposes an approach based on a state-space model for learning the user concepts in image retrieval.We first design a scheme galaxy harmony nc of region-based image representation based on concept units, which are integrated with different types of feature spaces and with different region scales of image segmentation.The design of the concept units aims at describing similar characteristics at a certain perspective among relevant images.We present the details of our proposed approach based on a state-space model for interactive image retrieval, including likelihood and transition models, and we also stickers logos dallas cowboys describe some experiments that show the efficacy of our proposed model.This work demonstrates the feasibility of using a state-space model to estimate the user intuition in image retrieval.