Software for Teaching and Learning

This demonstrates design, analysis and implementation issues associated with an orientation measurements system. The synthetic image below shows three 'bulbs' on the side of a 'space ship'. The mesaurement system is to compute the orientation of the space craft, for docking purposes, based on the image data. Unfortunately some impulse noise is corrupting the images and must first be removed via median filtering. An estimate of the accuracy of the orientation measurement is also needed, which is to be accomplished using a Monte Carlo Analysis.

Simulated image of three 'bulbs' , on the side of a 'space ship', which are used for an orientation measurement. The image is corrupted with impulse noise.

Monte Carlo is a simple and useful technique for studying the accuracy of a measurement system. Monte Carlo is very helpful when considering "what if…" scenarios during system design. This is especially so when processing includes nonlinear operations, which make it difficult to analyze via linear techniques (as with the median filtering, used here).

Median filtering is a type of neighborhood processing that is particularly useful for removing 'salt and pepper' noise from an image. Given a neighborhood defined as 3x3 or 5x5, for example, the filtered image is computed simply as the median value of the gray levels in each neighborhood region. (The median is the middle value of a sorted list).

The orientation measurement is found in a simplistic fashion here. Each illuminated pixel is assumed to be part of the bulbs, and each is used in the linear regression computation. This is of course problematic when the noise is present. (No blobs are found to isolate the three bulbs - see other demos).

A best-fit line is shown, as corrupted by noise .

Best fit line, after using a median filter to remove impulse noise.

The Monte Carlo method estimates accuracy via simulations. Because the input is simulated knowledge of the true orientation is available, for comparison with the measurements. This permits accuracy to be estimated by simulating many trials.

By using Monte Carlo Analyses repeatedly, studies of accuracy can be made for various system parameters. In the demo available below, the bulb size and bulb spacing may be adjusted. The demo also runs through trials with varying bulb radii and reports the estimated accuracy for each case. In this demo the Monte Carlo iterations are illustrated by updating the simulated images. This is for illustrative purposes.

Download Demonstration of Orientation Measurement System

Learning Objectives

The following learning objectives are associated with the above demonstration, with supporting lecture and possible programming assignment.