Research On Image Retrieval Method

With the rapid development of multimedia technology and Internet, image information resources retrieval has become the focus of research. Establishing effective image description and retrieval mechanism has become the problem which is needed to solve urgently. The most important problem of image retrieval is to solve the image semantic understanding.Spatial PACT (Principal component Analysis of Census Transform histograms) is a new representation for recognizing instances and categories of scenes. A spatial pyramid of PACT achieves higher accuracies.than the current state-of-the-art methods in several place and scene.The Bag of Words model is used is information retrieval. It will look a text as a collection of high-frequency words. Every word appears in the text is independent, not depend on whether there is the other word. Visual vocabulary model use different feature descriptors to extract feature for images, and then group the features into a specified number of clusters using k-means algorithm as feature descriptor for images retrieval.For the strength and efficiency of scene semantic recognition, we propose a new scene semantic recognition method based on Spatial PACT and Color feature. By means of importing potential step edge template into Spatial PACT, the efficiency of algorithm is improved greatly while the accuracy almost is not affected. At same time, by combining color feature, more powerful semantic representation is obtained. Experimental results show that the algorithm has high computational efficiency, high recognition rate and powerful semantic recognition.For the visual Bag of Words model, we propose an approach to retrieve images by means of a novel method named spatial visual vocabulary based on considering spatial similarity. Firstly, we hierarchically divide images into sub regions and construct the spatial visual vocabulary by grouping the low-level features collected from every corresponding spatial sub region into a specified number of clusters. To retrieve the image, the visual vocabulary distributions of all spatial sub regions are concatenated to form a global feature vector. Our method is a universal framework which is applicable to various types of features, so two kinds of features are used in the experiments. In almost all experimental cases, the proposed model achieves superior results. Our model, which works by dividing images into sub regions and computing spatial vocabulary among corresponding sub regions, has shown promising results on two large-scale, diverse datasets. It consistently achieves a significant improvement over classical bag of words model.DigiKam is an advanced digital photo management application. It provides a simple interface which makes importing and organizing digital photographs a "snap".DigiKam enables you to retrieval images using tags (keywords), captions, collections, dates, geolocation and searches. Therefore a powerful image retrieval system can be built by introducing characteristics descriptors to Digikam.