The Mathematical Visual Perception (MVP) Laboratory
Department of Applied Mathematics

Completed Projects:

##### Vision & Robotics

Sign Recognition

In this project a LEGO robot with a web cam was used. The robot moved on a scene including signs (such as traffic signs). Coming against a sign, the robot attempted to recognize the sign and act upon that recognition.

By: Guy Alon Zusman

Robot Homing

This project included a LEGO robot with a web cam mounted on it and an object called “home”. At the beginning of the experiment, we located the robot away from the home. The task of the robot was to search for the home, recognize it from other objects in the scene and guide towards the home in order to reach it (homing). These actions were based on the visual data captured by the camera.

By: Yaron Ben-Bassat

* Driving in Track

Motivated by the automatic car control problem, a robot with a web cam was constructed. Given a signed track, the task was guiding the robot to trace the track based on the visual input from the camera. The images captured by the camera were transferred to the host computer for image processing. Then proper steering commands were sent to the robot to follow the signed track.

By: Andrey Zhdanov & Dan Feldman

Stereo Vision Based on Variational Framework

In this project an implementation of Alvarez et al. (IJCV 2000) variational method for stereo vision is done.

By: Michael Medvinski

(Hebrew)

1)       Piano movers Problem

Here we used a LEGO robot in a scene which includes obstacles. A digital video camera was mounted above the scene. The robot mission included two tasks. The first task was calculating an optimal path for the robot based on the visual data captured by the camera. The second task was guiding the robot through the scene controlled by the visual data.

By: Tomer Livneh

Depth estimation from moving objects

In this project a sequence of images from a moving object were captured by a digital video camera. An algorithm was developed in order to estimate the depth map of the object from that sequence of images. Based on the depth estimation the 3D coordinates of every point on the object was evaluated. Then a 3D representation of the object was constructed. The moving object was a LEGO robot specifically constructed for this purpose and programmed to move at a constant speed.

By: Eric Rozenberg & Rona Oren

Final Report (Hebrew), Presentation, Movie (reconstruction)

##### Image Processing

Super Resolution

The goal is to use 2 or more images of an object with low resolution and reconstruct an image with higher resolution.

By: Alexey Ilyievski

Final Report (Hebrew),  Presentation (Hebrew)

-----------------------------------------------------------------------------------------------------------------------------------------

Remark: