Publication: Análisis del movimiento en secuencias de imágenes
Loading...
Official URL
Full text at PDC
Publication Date
2016
Authors
Advisors (or tutors)
Editors
Journal Title
Journal ISSN
Volume Title
Publisher
Citations
Exportar
Abstract
Este trabajo presenta el desarrollo de una aplicación destinada al análisis de secuencias de imágenes para la detección de movimiento en la escena. Se trata de un campo importante de la Visión Artificial, con múltiples aplicaciones entre las que se encuentra la videovigilancia con fines de seguridad, el control de tráfico, el movimiento de personas o el seguimiento y localización de objetos entre otras muchas. Para ello se utilizan métodos de análisis como son el de Lucas-Kanade y Gauss-Seidel, que obtienen el denominado flujo óptico. Este describe el movimiento que ha tenido lugar entre las imágenes y su fundamento estriba en la determinación de las variables espaciales y temporales en las imágenes, siendo precisamente la variable temporal la que introduce el concepto fundamental para el análisis del movimiento a partir de las imágenes captadas en diferentes instantes de tiempo dentro de la secuencia analizada.
Para el desarrollo de la aplicación se han utilizado técnicas propias del tratamiento de la Visión Artificial, así como la metodología proporcionada por la Ingeniería del Software. Así, se ha realizado una especificación de requisitos, se ha elaborado y seguido un plan de proyecto y se ha realizado un análisis de alto nivel, que se materializa en el correspondiente diseño e implementación, junto con las pruebas de verificación y validación, obviamente adaptados en todos los casos a las dimensiones del proyecto, pero que establecen claramente los planteamientos básicos para el desarrollo de una aplicación a nivel empresarial. La aplicación planteada se enmarca perfectamente dentro del paradigma, hoy en día en pleno auge, conocido como el Internet de las Cosas (IoT). El IoT permite la intercomunicación entre dispositivos remotos, de forma que mediante la correspondiente comunicación a través de conexiones a Internet es posible obtener datos remotos para su posterior análisis, bien en nodos locales o en la nube, como concepto íntimamente relacionado con el IoT. Este es el caso de la aplicación que se presenta, de suerte que los métodos de procesamiento de las imágenes pueden aplicarse localmente o bien transmitir las mismas para su procesamiento en nodos remotos.
This work presents the development of an application to analyze image sequences in order to detect movement in the scene. This is an important field in Computer Vision, with multiple applications such as video surveillance for security purposes, control of traffic and movement of people or tracking and localization of objects, among others. In order to do that, the Lucas-Kanade and Gauss-Seidel methods obtain the so called optic flow. It describes the movement that has taken place between the images, and its basis lies in the determination of the spatial and temporal variables in the images, being precisely the temporal variable the one that introduces the essential concept for motion analysis from the images captured at different times within the analyzed sequence. Typical techniques of Computer Vision have been used for the development of the application, as well as the methodology provided by the Software Engineering. Therefore a software requirements specification has been made, a project management plan has been developed and followed, and a high level design has been produced, which is materialized in the corresponding design and implementation, along with verification and validation tests. They are obviously adapted in all cases to the dimensions of the project, clearly establishing however the basic approaches to the development of an application on the business level. The considered application falls perfectly under the paradigm which is nowadays at its peak known as the Internet of Things (IoT). The IoT allows the intercommunication between remote devices, so that by means of the corresponding communications through the Internet it is possible to gather remote data for a later analysis, either in local nodes or in the cloud, as a concept closely related to the IoT. This is the case of the present application, so that the methods of image processing can be applied locally or rather broadcast the images to be processed on remote nodes.
This work presents the development of an application to analyze image sequences in order to detect movement in the scene. This is an important field in Computer Vision, with multiple applications such as video surveillance for security purposes, control of traffic and movement of people or tracking and localization of objects, among others. In order to do that, the Lucas-Kanade and Gauss-Seidel methods obtain the so called optic flow. It describes the movement that has taken place between the images, and its basis lies in the determination of the spatial and temporal variables in the images, being precisely the temporal variable the one that introduces the essential concept for motion analysis from the images captured at different times within the analyzed sequence. Typical techniques of Computer Vision have been used for the development of the application, as well as the methodology provided by the Software Engineering. Therefore a software requirements specification has been made, a project management plan has been developed and followed, and a high level design has been produced, which is materialized in the corresponding design and implementation, along with verification and validation tests. They are obviously adapted in all cases to the dimensions of the project, clearly establishing however the basic approaches to the development of an application on the business level. The considered application falls perfectly under the paradigm which is nowadays at its peak known as the Internet of Things (IoT). The IoT allows the intercommunication between remote devices, so that by means of the corresponding communications through the Internet it is possible to gather remote data for a later analysis, either in local nodes or in the cloud, as a concept closely related to the IoT. This is the case of the present application, so that the methods of image processing can be applied locally or rather broadcast the images to be processed on remote nodes.
Description
Trabajo de Fin de Grado en Ingeniería Informática y Matemáticas (Universidad Complutense, Facultad de Informática, curso 2015/2016)