Automated system for annotating and retrieving images

Automated annotation of digital images remains a highly challenging task being used for understanding of large collections of image data. Indexing and retrieval are better achieved when using the histogram as an efficient technique for content-based image retrieval domain. This paper presents an automated system that can be used for three distinct tasks: image annotation, semantic based image retrieval and content based image retrieval. The system is using an efficient annotation model called Cross Media Relevance Model for the annotation process. Image´s regions are described using a vocabulary of blobs generated from image features using the K-means clustering algorithm. Using SAIAPR TC-12 Dataset of annotated images it is estimated the joint probability of generating a word given the blobs in an image. Semantic based image retrieval is performed using two models provided by the annotation model. For content based image retrieval the system is using a combination of two types of histograms: a spatial color histogram and a texture histogram based on Local Binary Pattern descriptor. Both histograms are computed in the HVC color space. The computation of the binary code associated to a Local Binary Pattern descriptor is made using the NBS distance instead of using a simple difference between colors´ components.