Improving the Quality of Facial Image by Integrating Semantic Patches and Supervised Learning Approach

摘要

Image enhancement is the process of sharpening the image features by improving the quality besides reducing unwanted noise and blurredness. There are many software tools available for image enhancement, like different kinds of filters, editors, contrast enhancement, histogram equalization and many restoration methods for improving sharpness of an image. Image Enhancement is the route towards evolving an image with the objective that the outcome is more sensible for specific application. The proposed method suggested a new way to enhance an image and improve the quality by integrating non-rigid semantic patches technique with proposed machine learning algorithm. The primary goal is to construct the model or classify by training large set of images with prior information and consolidate semantic non-rigid patches from those images. Nearest neighbor classifier is used for identifying similar features, from the processed images. The proposed method is demonstrated for sample facial images and is suitable for application such as identifying criminal faces expressions or poses from the degraded and noisy images, in darker environments.