A New Content-Based Search Mechanism for Image Retrieval Search Engine

Introduction

In the present era of growing world of technology, the user wants the results of whatever user searches in a better and compact way. The Search Engines like Google, Bing etc. provide the facilities to search the information required using the three functionalities that are using text, using voice and also by using image. In the available Image Search Engines, when less than 50% part of image is entered as a query it is not able to retrieve the accurate result. Thus, the content-based image retrieval search engine is necessary. For searching about a monument or person or a place, it is necessary to know what it is called so that a text query can be formed. But in a case where one only has an image of the object/monument/person, it is difficult to form the query. The design of the search engine proposed here holds a possible solution to this problem. The objectives of the developing the search engine are: -

• •

  To create a dataset with different types of template patterns

• •

  To detect the level of accuracy in matching the image queried by the user

  • There are multiple CBIR techniques developed seeing the advancement in the image which helped in displaying the accurate output back to the user as expected. The techniques are mentioned below: -

    • o

      Query techniques based CBIR Retrieval

    • o

      Color based CBIR Retrieval

    • o

      Texture based CBIR Retrieval

    • o

      Shape based CBIR Retrieval

• •

  Query techniques based CBIR Retrieval: In this technique of query based CBIR Retrieval, the rough image is being passed to the system and then taking that as a form of example the algorithm applied in a top down manner to find out the resultant image related to the queried image.

• •

  Colour based CBIR Retrieval : Computing distance measure based on colour has always been a keen interest of development where the image retrieval mechanism completely focusses on the colour which is evident and unique in between multiple orientations of image. That means if the image is rotated also, the colour pixels will not change and will remain the same in order to form the picture. In this technique, all the results rely in developing the histogram, will be implemented and be a key component in order to identify and display the proper image back to the user precisely.

• •

  Texture based CBIR Retrieval: Texture measures identifies the visual patterns in the image and how they are spatially represented. Texture based retrieval are implemented by a number of combinational sets depending upon the count of textures being detected in the image. Texture is a property which allows the user to generate information about the spatial arrangement of color or intensities in an image or selected region of an image. Texture is the key ingredient used to identify objects or regions of interest in an image. And also, to identify that whether the image be a photomicrograph, an aerial photograph, or a satellite image.

• •

  Shape based CBIR Retrieval: This type of CBIR Retrieval technique is used in Edge detection of an image. Here, the Shape does not convey the shape of the image and it conveys the portion of image that is being sought out. This type of technique also focusses on the boundary of the image and is invariant to translation, rotation, and scale.