Research On Image Information Retrieval Based On Local Binary Patterns

The ever increasing amount of digital information requires effective information retrieval systems. As it is said, information which cannot be found easily is as good as lost. As information comes in various formats and types, their retrieval mechanisms also need to differ correspondingly. In this work, we deal with the task of content based image retrieval, in which the system facilitates the interaction between a user and an image database by automatic analysis of the image content.This thesis presents the research on Image information Retrieval based on Local binary patterns. Content-Based Image Retrieval (CBIR) has become a hot topic under image information retrieval in the recent time, under this thesis the basics of CBIR retrieval systems are analyzed where images are first characterized by descriptor vectors and retrieval is then based on these content-based descriptors. Selection of content based descriptors and defining suitable metrics are the core of any CBIR system. The local binary pattern (LBP) texture analysis descriptor operator have been defined as a gray-scale invariant texture measure, derived from a general definition of texture in a local neighbourhood. The LBP is proposed as a unifying texture model that describes the formation of a texture with micro-textons and their statistical placement rules.Measurement of image similarity is important for a number of image processing applications. Image similarity assessment is closely related to image quality assessment in that quality is based on the apparent differences between a degraded image and the original, unmodified image. Automated evaluation of image information retrieval systems relies on accurate quality measurement of similarity among the input image and the database images.This thesis discusses current algorithms for measuring similarity including mean squared error (MSE) and peak signal-to-noise ratio (PSNR). Due to their limitations such as: consistent, accuracy and greater computational cost, we introduced the local binary pattern operator where we treated the image under pixel level and compared different methods to get the best methods. The mean squared error (MSE) simulations have demonstrated its promise through a set of examples by showing its accuracy and low computational cost, though its performance is not good as compared to that of local binary patterns. We used different extension of local binary patterns to express our results, including block method and Circle texture method. The use of Local Binary patterns (LBP) features and the sub-block based image retrieval method which utilizes both global and local features was observed to be certainly more efficient than most of the existing retrieval methods that are based on global feature only. Experiment results show that this method can keep good translation, rotation, scale and image invariance and the retrieved results satisfy human visual perception remarkably. Compared to MSE results, our LBP operator is more robust than the previous methods and its required computation time is sufficient for image retrieval systems.