Clasificación de objetos en movimiento usando momentos de Jacobi-Fourier y la MTF del sistema óptico digital

A widely analysis for extracting features of moving objects using digital image processing techniques are presented. The propose method is based on the series expansion of the discrete image function in terms of the orthogonal circular polynomials such as Zernike, Mellin-Fourier, Chebyshev-Fourier a...

Descripción completa

Detalles Bibliográficos
Autor: CARINA TOXQUI QUITL
Tipo de recurso: tesis doctoral
Estado:Versión aceptada para publicación
Fecha de publicación:2010
País:México
Institución:Instituto Nacional de Astrofísica, Óptica y Electrónica
Repositorio:Repositorio Institucional del INAOE
Idioma:español
OAI Identifier:oai:inaoe.repositorioinstitucional.mx:1009/608
Acceso en línea:http://inaoe.repositorioinstitucional.mx/jspui/handle/1009/608
Access Level:acceso abierto
Palabra clave:info:eu-repo/classification/Reconocimiento de patrones/Pattern recognition
info:eu-repo/classification/Polinomios de Zernike/Zernike polynomials
info:eu-repo/classification/Función de transferencia óptica/Optical transfer function
info:eu-repo/classification/cti/1
info:eu-repo/classification/cti/22
info:eu-repo/classification/cti/2209
Descripción
Sumario:A widely analysis for extracting features of moving objects using digital image processing techniques are presented. The propose method is based on the series expansion of the discrete image function in terms of the orthogonal circular polynomials such as Zernike, Mellin-Fourier, Chebyshev-Fourier and other sets. This series expansion is supplied by a common mathematical tool, the generic function of orthogonal radial polynomials of Jacobi. This function provides a comparison of effectiveness of the polynomial families by means the variation of two parameters α and β which are used into its definition. A way to measure the performance of the polynomials studied here is by means the image reconstruction method. With this last technique the reconstructed image quality is evaluated respect to the original one, and at the same time the number of orthogonal moments required by the image reconstruction is determined. Typically, the moments employed in shape description and in moving analysis are divided into two categories: cartesian as the geometric moments and polar as the generic Jacobi moments. With this kind of polynomials, it can be generated a set of k invariant descriptors to rotation, scale, shifting, and intensity changes. Furthermore, it is implemented a discriminative measurement that is used to quantify the adaptation capacity of the polynomials to the difference between the shapes. The test images used in this thesis belong to objects with small intraclass variance and large interclass separation. With the help of optical-digital systems, different multidistortions are produced in the images during the acquisition and they are given by: 1) Geometric changes; as scale, shifting, and orientation in the vision field of objects, and 2) Blur by image motion; as linear, circular, and vibration in the cases of low and high frequency. In this work, a moving system is characterized by means the optical transfer function (OTF); which can be computed by the geometric moments of motion function of the object centroid. After the test images are acquired by the optical-digital system, their Jacobi-Fourier moments are computed and a vector of k features is generated for the reference and for the input objects. A selected features are determined by using the discriminative measurement. The results show that, with only one descriptor is possible to discriminate between different kinds of shapes.