손실 화소 추정을 이용한 고해상도 영상 개선에 관한 연구

Abstract

The image resolution enhancement is defined as one of the reconstruction techniques which enlarge the low resolution images to the high resolution ones. It is used as the basis technique of recognition and analysis of the objects not only in the fields of the general image processing but also in the area of specialization, such as satellite images and medical images. In this paper, the three approaches to the high resolution image enhancement which can generate the result images with better quality and low computational complexity are proposed. The aim is to develop the image resolution enhancement methods from the three following viewpoints: to enhance inexactitude of the object function, to alleviate uncertainty of the input low resolution, and to reduce the discordance with the original image. First, the method to improve the exactitude of the enhancement function is proposed to enhance inexactitude of object function. Next, the reflection-value estimation method is suggested to alleviate uncertainty of the low resolution image. Last, the missing-value estimation method used self-degraded restoration is proposed to reduce discordance with the original image. Through the experiment, it is proved that the proposed method in this paper is superior to any other conventional methods. It is also verified that the suggested method gets better results on each of the three standards for evaluation: the quantitative evaluation based on the objective index, the qualitative evaluation using the subjective criteria, and the evaluation on the computational complexity. The proposed method, which is characterized by independency, flexibility, expandability, and practicality of algorithms, is useful in the various applications where digital images are used.