An efficient region-based approach for object recognition and based on mathematical morphology and correlation coefficient

In this paper, a new technique based on slicing the images to equally sub-areas, then applying the density slicing to the color histogram of these areas combined with the color pair technique is presented as a filtration phase based on the color features of images. This allows dramatic reduction in the candidate sets of images, which needs to be proceed through the time-consuming process of shape feature extraction and comparison algorithms. After that, shape recognition method is proposed as a higher level phase in the proposed system. The proposed approach for shape-based retrieval relies on detecting the edges of the objects, using laplacian of gaussian, in the image query as well as the set of filtered images from the proceeding layer of color approach. Next, the morphological techniques, using dilate and erosion are applied, in order to solidify the objects within the image and to increase the chance of correlation coefficient to be used as means of similarity between images. Furthermore, transformation of objects is applied to overcome the rotation based problem. The experimental results show that the proposed approach is fast, efficient, and feasible.