Recognition of a pictured object in different images

Throughout the years there have been several different methods used for object recognition. However the methods used are far from providing satisfactory results for real time applications and for real environment images. Researches for accurate recognition of objects, in relation to their regional or general features, are currently being carried on.In this study, a methodology for the recognition of a pictured object in a different natural environment, regardless of its orientation, illimunation conditions, picturing angle, picture size, noise and distortion has been developed. The proposed methodology finds objects by initially using the object?s local features to match the similar points between the images, then searches similarities around the neighborhood of these matched points. The accuracy of the recognition, proved to attain a considerable increase, by means of the proposed methodology.