Integration of Color, Texture and Shape for Blood Cell Image Retrieval

The rapid development of technologies and steadily growing amounts of digital information highlight the need of developing an accessing system. Contentbased image indexing and retrieval has been one of the most important research areas in computer science for the last few decades. In this thesis, the approaches of contentbased image retrieval using low level features such as color, shape and texture are investigated to create a prototype that perceive images similar to a human. The color histograms and wavelet texture analysis in this thesis act as image based feature representations. The morphological operations like opening and closing are used to help analyze geometrical structures of medical images to extract shape as image components. In addition, the performance improvements on image retrieval offered by integrating the three promising features stated above. Experimental results using four different classes of 150 blood cell images showed 95.68% of retrieval accuracy.