Contenuto principale dell'articolo
Therefore, in the information searching it may result limitative to continue to operate in terms of a generic information retrieval. In the traditional practice, every kind of documental search is compelled to the conditions of a search through textual language; it is necessary, on the contrary, to consider a broader criterion of multimedia information retrieval, by which every kind of digital document is treated and searched through the elements of language more proper to its own nature. So, it is possible to differenziate, in a more general methodology of multimedia searching, a method of information retrieval based on textual information for the search of textual documents, from a method of visual retrieval based on visual data for the search of visual documents.
The main difference between the systems of storage and retrieval of textual documents and those of visual documents, is focused on the system of analysis and extraction of the elements representing the content of the document, and of the specific descriptors of its characteristics.
In the databases where the content of the documents is substantially textual, it is appropriate that the keys of access are terms and phrases extracted by the inside of that same content; in image databases, instead, it is inaccurate to attribute, from the outside, a textual description to contents that are well-grounded on a different structure of sense.
The new storage and retrieval technology treats directly the visual content of the documents, and for this it is defined content-based, in opposition to the traditional systems of indexing and searching based on terms describing such visual content, defined term-based. The method of the visual retrieval realizes, in conclusion, the possibility of searching images through the appropriate means of the visual language itself, as the similarity, the approximation and the relationships of measures and values, by using retrieval keys that are figures, textures, shapes, lines and colours.
In general, in the specialized systems of visual retrieval five different modalities are possible in order to index, storage, search and retrieve the visual documents, definable as ways of abstraction of the materials, that can constitute the structure of the system singularly or in different combinations among them.
Semantic modality is the most traditional method, but also the most problematic for images; it is founded on the definition of textual labels, describing characteristics, names, titles, classes or concepts, which must be known and recalled to allow the retrieval of the associated document.
Shape modality is based on the possibility to compare the shape extracted from the filed figure and the one extracted from the model on which the query is defined; the document retrieval may happen when the computer will appraise a certain degree of proximity among the values of the representative data of the compared images. The way of structural abstraction, instead, is founded on the decomposition of the images of the database in sections; the system will esteem the similarity of the structure of these with the structure of the sections of a sample figure, and it will retrieve the documents similar to such search keys.
Colour abstraction consists in representing images by extracting the different ranges of colours that constitute them; the operations of storage and retrieval will be based on the comparison of the values of the colour properties of the figures. Parametric modality, finally, is based upon the determination of the values of the indicative parameters of shape, texture and colour of the image; the system will effect the retrieval by comparing Whatever the modality of treating the documents is, and therefore of constructing the database, the images are analyzed and indexed at the moment of creating or updating the file, in manual way computer-assisted or in automatic way. In this system the search is, usually, planned starting from a first consultation of the database, almost always of semantic type, that allows to extract from it images that can become models, useful to launch the query in other forms, availing of a kind of visual thesaurus inside the database itself. The images drawn out each time can be modified, or selected by sections, relaunching different models that center the search in a better way.
More refined are the systems in which the images can be also analyzed in the phase of querying, and in which, so, the query models can be introduced by the outside, or designed with the provided tools, as finished sample images. The searches, in this case, can be conducted very freely, without the ties of an estabilished visual thesaurus. It is necessary, however, that the system technology is able, in every phase, to analyze and to elaborate automatically, and fastly, the external images proposed.
Many and different are the fields interested in the innovations of the visual retrieval, from the medicine to the geography, from the engineering to the visual arts, and each introduces very specific demands. The most important systems of search and retrieval, such as QBIC, VisualSEEk, ImageMiner or VIPER, have developed differently advanced solutions for every kind of problem.
Questo lavoro è fornito con la licenza Creative Commons Attribuzione - Condividi allo stesso modo 4.0.