Bibliographic citations
This is an automatically generated citacion. Modify it if you see fit
Camargo, E., (2016). Evaluación del desempeño de 2 métodos de súper resolución aplicado a imágenes médicas en 2D [Tesis, Universidad Nacional de Ingeniería]. http://hdl.handle.net/20.500.14076/5389
Camargo, E., Evaluación del desempeño de 2 métodos de súper resolución aplicado a imágenes médicas en 2D [Tesis]. : Universidad Nacional de Ingeniería; 2016. http://hdl.handle.net/20.500.14076/5389
@misc{renati/707779,
title = "Evaluación del desempeño de 2 métodos de súper resolución aplicado a imágenes médicas en 2D",
author = "Camargo Fernández Baca, Enzo",
publisher = "Universidad Nacional de Ingeniería",
year = "2016"
}
Title: Evaluación del desempeño de 2 métodos de súper resolución aplicado a imágenes médicas en 2D
Authors(s): Camargo Fernández Baca, Enzo
Advisor(s): Oporto Díaz, Samuel Alonso
Keywords: Súper resolución de imágenes; Procesamiento de imágenes; Calidad de una imagen digital: PSNR
Issue Date: 2016
Institution: Universidad Nacional de Ingeniería
Abstract: La presente tesis tiene como finalidad evaluar el desempeño de 2 métodos de Súper Resolución aplicados a imágenes médicas digitales en 2D: El método Iterative Back Projecting (IBP) y el método Iterative Kernel Steering (IKS), utilizando el indicador objetivo denominado PSNR (Peak Signal to Noise Ratio) para evaluar el desempeño de cada método en las pruebas sintéticas, el cual se calcula basado en las diferencias que existen entre la imagen original (HR) y la imagen en Súper Resolución (SR) obtenida.
La Súper Resolución es una tecnología diseñada en el campo del procesamiento de imágenes, diseñada con la finalidad de obtener una imagen de mayor resolución hasta 8 veces mayor a la imagen o imágenes de entrada, mediante el modelamiento de los fenómenos que afectan su obtención y minimizar sus efectos, mediante la adición de información que permita reconstruir la imagen real (HR) de forma más aproximada posible, logrando de esta manera la imagen en Súper Resolución (SR), esto es graficado en la Figura 27, donde se presenta el Esquema general de la Tesis, presentando el problema y su solución mediante Súper Resolución.
En la presente tesis se logra mejorar la calidad de la imagen hasta 3 veces (3x) en relación a la imagen inicial obtenida, es decir, si la imagen de entrada tiene una resolución de 240 x 240 pixeles, se incrementa hasta en 720 x 720 pixeles, 3 veces en cada dirección ancho y alto.
En relación al desempeño obtenido de métodos evaluados se demuestra que el método IBP (varias imágenes de entrada) es mejor el IKS (1 sola imagen de entrada), en los que no hay presencia de ruido, es decir, PSNR (IBP) > PSNR (IKS), asimismo hay casos que los que predomina el ruido, se observa que el IKS tiene mejores resultados obteniéndose PSNR (IBP) < PSNR (IKS).
El aporte del presente trabajo y su desarrollo es trascendental por las potenciales aplicaciones futuras en mamografía, ecografía, rayos X, digitalización de placas de imágenes médicas, así como aplicaciones en la industria, seguridad civil y militar, exploración minera, forestal, hídrica y recursos en general, esto debido a que las imágenes al ser digitalizadas se pueden procesar y explotar de muchas maneras y extraer información relevante según su área de aplicación para la toma de decisiones.
Palabras Clave: Súper Resolución, Método Iterative Back Projecting (IBP), Método Iterative Kernel Steering (IKS), mejora de la resolución de una imagen digital, fenómenos que afectan la resolución de la imagen, ruido, sub muestreo, desenfoque, interpolación de imágenes, Pico de Señal Ratio– Ruido o Peak Signal to Noise Ratio en ingles (PSNR), Punto de Funcion de Propagación o Point Spread Function (PSF), Pruebas sinteticas, imágenes médicas digitales, aplicaciones para el desarrollo nacional, aplicaciones en imágenes médicas, aplicaciones en imágenes satelitales.
Digital images have been widely applied in medicine like X rays, ecography, MRI among others. However, the image analisys can be limited by motion blur, movement induced by patient behaviour such little kids, low performance of equipment and the noise inherent within the image sensors, that every image digital equipment has. Therefore, the improvement of image quality becomes an important task for image analisys. I present super resolution methods as a important solution for improvement of image quality and finally evaluate their performance in medical images, also with photos. Super resolution is the process in which one or more images are used for the construction of just one superior resolution image more resolution than the input images.The input images are known as a low resolution image (LR: Low resolution image) and the result image are known as super resolution image (SR: Super resolution image). The applications of this technology are many besides medical are military, security, robotics, UAVs among others. NASA also uses super resolution for the image obtained from telescopes, satelites and in the aerospace research. Peru has realized important advances in the aerospace field, UNI has launched its own satellite and it was the first satellite launched by our Country history. Therefore this field of image processing and super resolution is so important for national development. Here, we develop a super resolution framework, one method is based on frecuency domain, uses iterative back projecting to achieve the super resolution image using many input images; and the another is based on methods that need just one image. A quantity comparison in computation time, visual performance and PSNR for measure objectively the image quality are performed. The results shows a great improvement of the resolution in 3x, for example one image of 171x171 pixeles is improved to 512x512 pixeles and the low resolution images of 30.19 dB for synthetic images, and a considerable visual improvement in sharpness and visual details for real medical images and real pictures images. The method based on many input images are more efficient and get better results than the method based on just one input image. Keywords: Super resolution, Iterative back projecting method (IBP), Iterative Kernel Steering, improve of quality digital image, phenomena that affects image resolution, noise, aliasing, downsampling, upsampling, interpolation, peak signal to noise ratio (PSNR), PSF (Point Spread Function), synthetic test, digital imaging, applications for national development, medical imaging, satelital imaging.
Digital images have been widely applied in medicine like X rays, ecography, MRI among others. However, the image analisys can be limited by motion blur, movement induced by patient behaviour such little kids, low performance of equipment and the noise inherent within the image sensors, that every image digital equipment has. Therefore, the improvement of image quality becomes an important task for image analisys. I present super resolution methods as a important solution for improvement of image quality and finally evaluate their performance in medical images, also with photos. Super resolution is the process in which one or more images are used for the construction of just one superior resolution image more resolution than the input images.The input images are known as a low resolution image (LR: Low resolution image) and the result image are known as super resolution image (SR: Super resolution image). The applications of this technology are many besides medical are military, security, robotics, UAVs among others. NASA also uses super resolution for the image obtained from telescopes, satelites and in the aerospace research. Peru has realized important advances in the aerospace field, UNI has launched its own satellite and it was the first satellite launched by our Country history. Therefore this field of image processing and super resolution is so important for national development. Here, we develop a super resolution framework, one method is based on frecuency domain, uses iterative back projecting to achieve the super resolution image using many input images; and the another is based on methods that need just one image. A quantity comparison in computation time, visual performance and PSNR for measure objectively the image quality are performed. The results shows a great improvement of the resolution in 3x, for example one image of 171x171 pixeles is improved to 512x512 pixeles and the low resolution images of 30.19 dB for synthetic images, and a considerable visual improvement in sharpness and visual details for real medical images and real pictures images. The method based on many input images are more efficient and get better results than the method based on just one input image. Keywords: Super resolution, Iterative back projecting method (IBP), Iterative Kernel Steering, improve of quality digital image, phenomena that affects image resolution, noise, aliasing, downsampling, upsampling, interpolation, peak signal to noise ratio (PSNR), PSF (Point Spread Function), synthetic test, digital imaging, applications for national development, medical imaging, satelital imaging.
Link to repository: http://hdl.handle.net/20.500.14076/5389
Discipline: Ingeniería de Sistemas
Grade or title grantor: Universidad Nacional de Ingeniería. Facultad de Ingeniería Industrial y de Sistemas
Grade or title: Ingeniero de Sistemas
Register date: 13-Oct-2017
This item is licensed under a Creative Commons License