Contour Shape Descriptor for 2D Image Retrieval

Conference: Creative Trends in Engineering and Technology
Author(s): Pushpalatha S.Nikkam, Priyanka Mathapati, Pratibha Badiger, Nagaratna P. Hegde, B. Eswara Reddy Year: 2016
Grenze ID: 02.CTET.2016.1.522 Page: 196-208

Abstract

Recognizing contour of any object in a digital image is a challenging task. This paper presents an efficient method\nto identify the contour of an object in a given image and retrieve similar images. Initially major noise in an image is removed\nusing bilateral filters and then threshold is applied to eliminate small noises. Canny Edge detection technique is used to get\nthe edges in an object. The morphological dilation, erode and flood fill techniques are used to draw the contour of the object.\nThe shape features are computed using the area, bounding box, perimeter, and centroid of the contoured image. Shape based\nimage retrieval depends on measuring of similarity between shapes represented by their features. The distance measures such\nas Euclidean, Canberra, Manhattan are used to retrieve similar image from the database. The experimental results are\ntabulated using precision and recall graph.

<< BACK

CTET - 2016