A Survey on Image Feature Selection Techniques

Image is a work of art that describes or store visual perception. That has an exactly same appearance to subject – normally a physical object or a person. Therefore, images providing a representation of real time physical objects. Additionally, representation of 2 dimensional data and their patterns analysis is basic interest of the given paper. As a part of object recognition, the image and their objects can be recognised using their pattern. But there are some issues are involved for finding the essential patterns from image data. First, the amount of data to be analysed, Secondly, a simple and single class may have a huge amount of data to be process due to their class definition. In this context, dimensionality reduction techniques are helpful for reducing amount of data and finding the accurate pattern form the image data. Therefore, in this paper PCA, LDA and ICA methods are studied. These techniques are the most popular and frequently used method of pattern extraction from different kind of data types i.e. image and huge datasets. In addition of that, these techniques are helpful in classifying the objects based on their extracted pattern. The presented paper introduces discussion about these techniques. Keywords— PCA, LDA, ICA, Pattern extraction, feature