The color face hallucination using Eigentransformation with error regression model

Abstract

This paper proposes a new face hallucination technique for color face image reconstruction using Eigentransformation with error regression model. Generally, a high-resolution (HR) face image is reconstructed only from low-resolution (LR) face image. Nevertheless, previous researches neglect to gain benefit from error of face reconstruction. Therefore, in order to improve the performance of facial image reconstruction, the error information is included in our framework to correct the final result. In this way, the regression analysis was used to find the error estimation, which can be obtained from the existing LR in eigen space for each color channel. To handle with color image, each color channel in RGB model is separately processed by our framework. Our framework consists of two-phase. First phase learn error of face image reconstruction from training dataset by Eigentransfomation, then regression analysis is performed to find relationship between input and error. Second phase hallucinates using normal method by Eigentransformation, after that the result is corrected with the error estimation. Experimental results on the well-known database show that the resolution and quality of the hallucinated color face images are greatly enhanced over the traditional Eigentransformation method, which is very helpful for face recognition.