Fuzzy Object Shape for Image Retrieval

Compared to color and texture, the shape is considered as an important feature for many real-time applications. In this chapter, Fuzzy Object Shape (FOS) is presented for extracting the shape information present in the images. It is further noticed that the boundary of the object is ill-defined and there is impreciseness and vagueness in the object information. The closeness of the object with well-known primitive shapes are estimated. It is known that the impreciseness can be effectively captured by fuzzy functions and FOS has offered seven fuzzy membership function for the same. The value of each fuzzy membership function are constructed as feature vector to define the properties of individual objects.

Textual-Shape-Based Image Retrieval

In this chapter, a method to combine both text and image feature is considered. The FOS is explained in Chapter 3 is combined with textual information extracted (as discussed in Chapter 1). A clustering mechanism is formulated based on image, text and both. A retrieval is presented as an example to demonstrate the functionality by which the reader can understand the use of combining both textual keywords and FOS. The Chapter has consolidated the performance of combined feature using Precision, Recall and F1-score. The performance is evaluated and compared with well-known Google retrieved system.

Encoding and Indexing Fuzzy Object Shape for Image Retrieval

The recent retrieval and indexing approaches suffer from the issues of curse of dimensionality, overlapping of vectors, need of extra parameters for clustering and not supporting incremental indexing. In this chapter, an indexing approach is introduced without enduring the above specified issues. The suggested indexing structure is dynamically rearranged based on the occurrences of pattern for merging the common patterns in low-level feature. Further, the feature is encoded using GR coding scheme and the feature database space required is reduced considerably. While there are many encoding scheme available, in this chapter GR coding is used for simplicity and its applicability. In addition, the compression ratio is discussed and numerical statistics is depicted. Overall, indexing scheme reduces the search space by following predetermined pattern for clustering. The encoding scheme reduces the size of the feature database and also achieves good precision of retrieval.

Text-Based Image Retrieval

The role of textual keywords for capturing the high-level semantics of an image in HTML document is studied. It is observed that the keywords present in HTML documents can be effectively used for describing the high-level semantics of the images appear in the same document. Techniques for processing HTML documents and Tag Ranking for Image Retrieval (TRIR) is explained for capturing semantic information about the images for retrieval applications. A retrieval system returns a large number of images for a query and hence it is difficult to display the most relevant images in top results. This chapter presents newly developed method for ranking the images in Web documents based on the properties of HTML TAGS in web documents for image retrieval from WWW.

Similarity Measure for Matching Fuzzy Object Shapes

In this chapter, the Common Bin Similarity Measure (CBSM) is introduced to estimate the degree of overlapping between the query and the database objects. All available similarity measures fail to handle the problem of Integrated Region Matching (IRM). The technical procedure followed for extracting the objects from images is well defined with an example. The performance of CBSM is compared with well-known methods and the results are given. The effect of IRM with CBSM is also proved by the experimental results. In addition, the performance of CBSM in encoded feature is compared with similar approaches. Overall, the CBSM is a novel idea and very much suitable for matching objects and ranking on their similarities.

Content-Based Image Retrieval Using Shape Features

This chapter presents CBIR methodologies for extracting geometric and margin features of objects in images and constructed as feature vector. This approach is unique in nature as the size of the feature is relatively small and capable of discriminating the query object with the data base object. These geometric features measure the object characteristics in terms of its shape and margin. Manhattan distance is used for measuring the similarity between query images and the database images for retrieving relevant images from the database.