Desarrollo de técnicas de inpainting y demosaicing basados en el modelamiento disperso
Portada
Citas bibliográficas
Código QR
Autores
Director
Autor corporativo
Recolector de datos
Otros/Desconocido
Director audiovisual
Editor/Compilador
Editores
Fecha
Cita bibliográfica
Título de serie/ reporte/ volumen/ colección
Es Parte de
Resumen
One of the most important tasks of image processing in photography is to improve the quality of the photographs you want to study. Many times this is difficult if the acquired image contains scratches, patches or some object or drawing that obstructs the desired information, losing important information that may be needed. The method used to eliminate these possible obstacles thus improving the quality of the image is call inpainting.Another important problem is to treat the reconstruction of the image from the pixels acquired by the digital camera, since cellular cameras, for example, do not acquire the entire pixel matrix and a filling process called demosaicing is necessary, which seeks to complete the image with the least possible distortion. Recently, techniques based on scattered models appeared that offer a solution to the problems described above and allow their implementation in advanced research. However, these techniques are complex to use. Therefore, this paper presents a detailed analysis of the methods and their implementation in a free software ImageJ, also allowing its use in microscopy and biology. The model is a system of linear equations that describes the use of a dictionary next to a vector to make the reconstruction or improvement of the images. In this work, techniques such as OMP (Orthogonal pairing search) will be developed to make the dictionary sweep, decompose the image into patches and make the writing of it. In addition, a dictionary search technique, known as K-SVD, was developed for better results