This project aims to develop a software to do a searching process based on the content of the images rather than their textual description. In general, searching for some content-specified image is a big delimma because the images are picked up according to their description rather than their pictorial content. The ContentSpecified Image Searcher software will adopt a strategy to perform this searching process based on searching for an image according to at least two selected sample images by the searcher. The searcher will be required to enter at least two keywords and a list of sample images will be displayed for each keyword and the searcher is required to select only one sample image from each list.
The searching process will be now performed with reference to the sample images selected by the searcher. The opportunity to find the desired image increases as many as sample images are selected. For example, if the searcher looks for an image representing an open book and a handed len magnifying the text on the open page, the searcher should select a sample image for an open book and a sample image for a len. Accordingly, the searcher will get a pack of results one of which may be the desired image. If the seacher wants the book to be antiquarian, s/he should select a third sample image for an antiquarian in addition to the sample images of the open book and the len, and so on.
