Image Retrieval Based on Content Using Color Feature

Algorithm 2

The proposed CBIR algorithm.

Purpose: The algorithm is to retrieve images similar to the input image.

Input: An RGB image, number of retrieved images n.

Output:  n images similar to the input image.

Method:

Step 1 : The input image is a color image in RGB color space.

Step 2 : Apply the Ranklet Transform for each image layer (R, G, and B). The output images will be in three orientations (vertical,

horizontal, and diagonal).

Step 3 : For each ranklet image (vertical, horizontal, and diagonal) in a specified layer, calculate the color moments (8),

(9), and (10).

Step 4 : Construct the feature vector that will represent the image containing 27 numerical values.

Step 5 : Cluster the images in the database using k-means algorithm (Algorithm 1) into different categories.

Step 6 : Calculate the distance between the input image and the centroid of each cluster using Euclidian Distance, and find the

smallest distance.

Step 7 : Calculate the distance between the input image and the images in the cluster that has the smallest distance with the

input image.

Step 8 : Retrieve the first n images that is similar to the input image.