Search

During the implementation of computer vision algorithms the manipulation of pointers, memory administration and some other resources are expensive in time even for friendly programming language. All these problems can be resolved if the implementation test is carried out in MatLAB using its toolbox of image processing with it the time of implementation becomes the minimum with the trust of using algorithms scientifically proven and robust. In this work we show the form in which can be used matlab and its toolboxes to solve common problems of computer vision efficiently.

The Kalman filter has been used successfully in different prediction applications or state determination of a system. One important field in computer vision is the object tracking. Different movement conditions and occlusions can hinder the vision tracking of an object. In this report we present the use of the Kalman filter in the vision tracking. We consider the capacity of the Kalman filter to allow small occlusions and also the use of the extended Kalman filter (EKF) to model complex movements of objects.

El filtro de Kalman ha sido usado exitosamente en diferentes aplicaciones de predicción o determinación del estado de un sistema. Un campo importante en la visión por computadora es el seguimiento de objetos. Diferentes tipos de movimiento así como el ocultamiento de objetos a seguir pueden obstaculizar la labor de seguimiento. En este trabajo, se presenta el uso del filtro de Kalman para el seguimiento de objetos. Además se considera tanto la capacidad del filtro para tolerar pequeños ocultamientos del objeto así como el uso del filtro de Kalman extendido para el modelaje de movimientos complejos.

This work report the electronic and mechanical for a bipedal robot prototype. The objective of this project is build a platform to study the dynamic walking and to prove some intelligent control techniques. This work highlighted the fact that mechanical design is equally, if not more important, than the control method used. All the elec- tronic was implemented in a low cost PIC16F873 microcontroller. The robot designed has 9 degree of freedom (DOF) and each joint is driven by a DC servo motor. A gait program besides a control algorithm who read a gyroscope and force sensors are used to equilibrate the robot at walking.

The segmentation of objects whose color-composition is not trivial represents a difficult task, due to the illumination and the appropriate threshold selection for each one of the object color-components. In this work we propose the Fuzzy C-Means algorithm application for the segmentation of such objects. It is chosen, by the characteristics that it represents the face segmentation. This technical report is organized in the following way: in section 1 a clustering techniques introduction are presented. In section 2 Fuzzy C-means algorithm is analysed and also showed with a simple example. In section 3 Matlab tools, that are used to code the fuzzy C-means algorithm are described. In section 4 the Fuzzy C-Means algorithm is implemented for the face segmentation. Finally in section 5 the results are presented and the possible improvements are proposed.

In the robotics area, visual tracking is an important and difficult problem therefore is necessary to have a robust and efficient control algorithm which presents immunity characteristics to stochastic direction and speed changes of the object to be tracked. Also is important count with a segmentation algorithm which be able to tolerate changes in the intensity of light. We describe in this report the implementation of fuzzy controllers based on the fuzzy condensed algorithm and also the developed of a LVQ neural network to segment the image. For this work we used two fuzzy condensed algorithms running in a PC to control a robot’s head which tracks a human face. We describe the main lines of the fuzzy condensed algorithm as well as the LVQ neural networks architecture employed and the implementation, the fuzzy condensed controller performance in comparison to a PID controller and real time results.

Color-segmentation is very sensitive to changes in the intensity of light. Many algorithms do not tolerate variations in color hue which correspond, in fact, to the same object. Learning Vector Quantization (LVQ) networks learn to recognize groups of similar input vectors in such a way that neurons physically near to each other in the neuron layer respond to similar input vectors. Learning is supervised, the inputs vectors into target classes are chosen by the user. In this work a new algorithm based on LVQ is presented. It involves neural networks that operate directly on the image pixels with a decision function. This algorithm has been applied to spotting and tracking human faces, and shows more robustness than other algorithms for the same task.

In the robotics area, visual tracking is an important and difficult problem therefore is necessary to have a robust and efficient control algorithm which presents immunity characteristics to stochastic direction and speed changes of the object to be tracked. Also is important count with a segmentation algorithm which be able to tolerate changes in the intensity of light. We describe in this report the implementation of fuzzy controllers based on the fuzzy condensed algorithm and also the developed of a LVQ neural network to segment the image. For this work we used two fuzzy condensed algorithms running in a PC to control a robot’s head which tracks a human face. We describe the main lines of the fuzzy condensed algorithm as well as the LVQ neural networks architecture employed and the implementation, the fuzzy condensed controller performance in comparison to a PID controller and real time results.