Image retrieval

Published on 26 Jun 2021. Written by Akshay

Due to the emerging technologies and advancements in the growth of computers the data and the information are growing day by day.

Read more..

The maintenance of the existing data in computers is very hard since there is a lot of information such as images, files, etc.

The problem now arising is that when a user searches some data on the internet, they are not able to find the appropriate data and hence the work gets late.

Hence to avoid similar kind of problems many systems were developed such as text-based image retrieval (TBIR) system, which uses google images

Since the data is too large so there should be more development and features like searching for image based on properties such as color, shape, the texture should be available

Hence after many kinds of research, a new method content-based image retrieval (CBIR) method was developed to sort the huge data based on its color, shape, and texture.

Content-Based Image Retrieval (CBIR): It is a technique that sorts the large scale database based on images visual content that is color, texture, and shape and gives the output of user request in the form of an image.

In this project, the color retrieval is done by HSV color space based color histogram and the image retrieval is done by Gray Level Co-occurrence Matrix (GLCM)

Canberra distance is used for this retrieval feature.

The colors which are used by computers, graphic card, etc. are the RGB color space. The primary colors are red, green and blue.

Color feature extraction

The following steps are carried in color feature extraction

• The user provides the image query, make a program to read it
• The given format will be in RGB which is to be converted to HSV
• The space of HSV should be quantized to 256 histogram bins
• The 256 histogram bins values that are present to be stored as a color feature in the database of vector feature
• Now Canberra distance is used to measure the similarity and the images are retrieved based on the minimum distance

Shape feature extraction

The following steps are carried in shape feature extraction

• The user provides the image query, make a program to read it
• The given format will be in RGB which is to be converted to a greyscale image
• Calculate 4 gradients of morphology and 7 variants of moments for each edge map
• Compare the received database image to the query image and retrieve the images based on the minimum distance

 Texture feature extraction

The following steps are carried in texture feature extraction

• The user provides the image query, make a program to read it
• The given format will be in RGB which is to be converted to a greyscale image
• Calculate 4 GLCM matrix containing energy features, homogeneity features, contrast features, and co-relation features
• Compare the received database image to the query image and retrieve the images based on the minimum distance

Kit required to develop Image retrieval:

• NA

Technologies you will learn by working on Image retrieval:

• NA

Any Questions?