Scalable Face Image Retrieval using Attribute-Enhanced Sparse Code words

Scalable Face Image Retrieval using Attribute-Enhanced Sparse Code words

People like having and saving photos every time, with friends, family, superstars etc. with each passing day the technology is advancing so is the photo quality and the cameras. Thus, with the progression and development in photo capturing devices, photo quality is also exponentially growing. Large scale content based image retrieval is an empowering technology in developing software’s and applications. This article has aimed to address the main problem of large scale content based image retrieval. This empowering technology enables the users to find the similar and comparable images of a single image in a large database storage system. This content based image retrieval is a quite useful technology which will help in crime investigations, automatic face annotation, etc.

In the traditional system, method used for the image retrieval typically uses the low image features of a face or appearance to find other similar faces and as a result of which, most of the tomes lack semantic features of a face and the relatable images found have a wide range of difference between them. Thus, the results of this kind of image retrieval were unsatisfactory.

This research work has targeted and established to deliver a new perspective to the image retrieval system by integrating high-level human qualities and characteristics into face image representation and index construction. People may resemble and look alike if we compare their low level features but, by combining and comparing the high level attributes of faces we will be able to find better retrieval results and improved representation of similar faces.

In this research two potential methods has been proposed which will provide the most desired and provide searches by analyzing the fine attributes. These attributes will be evaluated and analyzed with the help of the attribute detectors to enhance the results and advance the system of the image retrieval based on the content based face images

In this study, it has been established that the automatically detected images based on the attributes also contain and show some of the semantic indications and prompts. It will develop that the face photos will now be constructing and will be retrieved on semantic cue too.  In this study the two methods have been proposed, these methods are known as the attribute embedded inverted indexing and attribute-enhanced sparse coding. These methods have been proposed and suggested to advance the face retrieval method and system.  The efficiency of different characteristics and vital factors indispensable for face repossession has been recognized in this research study.

Advantages:

To sum up, the contributions of this paper include:

This study has collectively and automatically identified and used low and high level features of the face and it attributes, besides, this study has also combined the semantic cues. This is very basic study in this regard to enhance the system of the image retrieval based on the content.

This content based retrieval of the images during this study has also identified that generic and informative human attributes can also be used to retrieve images and can be utilized in different data bases.

Advantage of this study is that it has combine two different orthogonal approaches and improved the system of the image retrieval.

In the literature review search it has been revealed that the system is the very basic and first of its kind and no system has been developed to detect the content based images by using low level feature and face attributes.

It has also been also discovered that the certain informative attributes for face retrieval and this whole system is quite useful for other purposes also such as face recognition and detection.

Current prevailing methods t and systems treat all attributes as corresponding and identical.  This research has established the methods to vigorously decide the standing of the characteristics and supplementary abuse of the contextual dealings and associations between them.