Description

Retrieval of similar images from database has numerous applications in mining process. Manually tagging images is both subjective and time consuming. It would be nice if the tagging process could be done automatically. A requirement for effective searching and retrieval of images in rapid growing online image databases is that each image has accurate and useful annotation. Manually tagging images is both time consuming and subjective. People simply do not bother or have time to tag their images. Furthermore, human beings are and think differently, meaning that similar images will be tagged differently by different people. A tag is a keyword or term assigned to an image that helps describe the image and its content so that it can easily be retrieved when searching or browsing for it. Image tagging is the process of assigning tags to an image. Image annotation is the process of annotating images with relevant information. The identification of the related tags for the input images helps in providing more accurate mining results. But almost all kind of information can be used in some way. The key is to use the information that will assist most in achieving the desired goal. For images, it could be important to know where, when and in which situation the image was taken or what the main subject of the image is. The main reason to tag images is for improving the usability of image retrieval systems. Image retrieval systems can be used to find images of interest in both large online public image databases and personal image collections stored on home computers. A process that retrieves the similar tags from the image based on the texture based features extracted from the image is proposed. The texture based features were extracted from the images based on Histogram of Gradients (HOG) algorithm is proposed. The similarity in the features were identified based on the distance measurements based on Euclidean distance. The visual features extracted from the images were helpful in the identification of similar tags in the images.