Image Retrieval: Fundamentals and Applications

By Fouad Sabry

What Is Image Retrieval

A computer system that is used for browsing, searching, and retrieving images from a vast collection of digital images is called an image retrieval system (sometimes abbreviated as IRMS). In order for image retrieval to be carried out over the annotation words, the majority of the conventional and widespread methods currently in use include the addition of information to the images themselves. This metadata can take the form of captioning, keywords, titles, or descriptions. Annotating images manually is a significant investment of time, effort, and money; as a result, a significant amount of effort and research has been put into developing automatic image annotation methods. In addition, the proliferation of social web apps as well as the semantic web has been a driving force behind the development of a number of picture annotation tools that are web-based.

How You Will Benefit

(I) Insights, and validations about the following topics:

Chapter 1: Image retrieval

Chapter 2: Information retrieval

Chapter 3: MPEG-7

Chapter 4: Content-based image retrieval

Chapter 5: Automatic image annotation

Chapter 6: Image organizer

Chapter 7: Google Images

Chapter 8: Image meta search

Chapter 9: Metadata

Chapter 10: Reverse image search

(II) Answering the public top questions about image retrieval.

(III) Real world examples for the usage of image retrieval in many fields.

(IV) 17 appendices to explain, briefly, 266 emerging technologies in each industry to have 360-degree full understanding of image retrieval' technologies.

Who This Book Is For

Professionals, undergraduate and graduate students, enthusiasts, hobbyists, and those who want to go beyond basic knowledge or information for any kind of image retrieval.

• Language: English
• Publisher: One Billion Knowledgeable
• Release date: Jul 6, 2023

Book preview

Chapter 1: Image retrieval

To view, search for, and retrieve digital images from a large database, you need an image retrieval system. The majority of common and established approaches to retrieving images rely on annotating them with metadata like captions, keywords, titles, and descriptions. Since manual image annotation is a costly, time-consuming process, there has been extensive study into automating the process. Several web-based image annotation tools have also been developed as a direct result of the rise of social web applications and the semantic web.

Banireddy Prasaad, Amar Gupta, Hoo-min Toong, and Stuart Madnick of MIT created the first ever microcomputer-based image database retrieval system in the 1990s.

As of the year 2021, no image retrieval systems were built for 3D images at all, only 2D ones.

To find images, you can use a data search technique called image search. A user can perform an image search by entering a keyword, selecting an image file/link, or clicking on an image; the system will then return results that are similar to the user's selection. Meta tags, color distribution in images, region/shape attributes, etc. could all serve as the similarity used for search criteria.

Images can be searched for based on their metadata, which can include keywords, descriptions, and more.

The use of computer vision in image retrieval, also known as content-based image retrieval (CBIR). The goal of content-based image retrieval (CBIR) is to eliminate the need for textual descriptions by retrieving images that share visual characteristics (textures, colors, shapes, etc.) with a query image or user-specified image features.

Search engines that index results based on visual characteristics of images, such as color, texture, shape/object, and so on are included in the CBIR Engines List.

Searching for images in a collection by employing unconventional exploration strategies.

In order to gauge the difficulty of designing an image search system, it is critical to grasp the breadth and depth of image data. Search engine designs are heavily impacted by factors like the anticipated number of users and the demographics of those users. Along this axis, we can categorize search results as follows::

Archives - typically store large amounts of topic-specific, homogenous data in a structured or semi-structured format.

Domain-Specific Collections are those that are both homogeneous and accessible only to authorized users working toward a narrow set of goals. Databases in the fields of medicine and satellite imagery are two types of such collections.

Enterprise Collection refers to an organization's internal image database, which contains a wide variety of file types. There are numerous options for archiving photographs.

A personal collection is one that is primarily accessible by its owner, is compact in size, and is typically stored locally on a storage medium.

Images posted to the World Wide Web can be viewed by anyone with access to the World Wide Web. Large disk arrays are typically used to store these semi-organized, heterogeneous, and massive image collections.

Workshops dedicated to assessing and enhancing the functionality of image retrieval systems are held regularly.

ImageCLEF is an ongoing sub-forum of the Cross Language Evaluation Forum that compares and contrasts image retrieval techniques with textual ones.

From 1998 to 2001, the IEEE hosted a workshop series titled Content-based Access of Image and Video Libraries..

{End Chapter 1}

Chapter 2: Information retrieval

Computer scientists and librarians define information retrieval (IR) as the process of retrieving from a collection of information system resources that are relevant to an information demand. Full-text indexing and other forms of content indexing may be used to conduct searches. The field of study known as information retrieval focuses on the processes involved in finding specific pieces of data among large collections of text, pictures, and audio files.

The term information overload describes the problem that automated information retrieval systems aim to solve. Access to books, journals, and other publications is only the beginning of what an IR system can do for you. The most well-known IR applications are web search engines.

When a user or searcher inputs a query into the system, the process of retrieving the requested information starts. Queries are structured expressions of information demands, such the search strings used in online search engines. A query in information retrieval does not always result in a uniquely