Sketch Based Image Retrieval with Cosine Similarity

Main Article Content

Y. Jhansi
Dr. E. Sreenivasa Reddy

Abstract

In the present paper cosine similarity is used for similarity measurement in sketch to image retrieval with focus on image matching. Similarity measurement is very challenging task in sketches due to its variable dimensions in the edges. Median filters are used to minimize the noises occur in the images due to unknown reasons. After pre-processing of the images in the benchmark database, the features are extracted using Histogram of oriented gradients (HOG). Canny edge detection algorithm is discussed to detect the image edges in several directions. Later the sketch to image matching is calculated by similarity measurement algorithm cosine similarity. A Benchmark sketch image database is considered to prove the proposed algorithms is efficient and works effectively when compared with many existing techniques. It is very important to measure accurate similarity measurement in high dimensional space for information retrieving. This paper presents efficient cosine similarity measurement algorithm. In the present exertion extensive experimental work is carried out inorder to prove the state of art in association with Sketch Based Image Retrieval (SBIR). Keywords: Image retrieval, Cosine similarity, HOG, Canny edge detection, Median filter

Downloads

Download data is not yet available.

Article Details

Section
Articles