Heterogeneous Data Fusion
Homework 1
Due February 18th, 2015 – 7:00 P.M.
Submit your completed homework and associated models via Blackboard.
1. A computer vision system must distinguish between Rubronian tanks, armored personnel carriers, trucks, and mobile missile launchers. The system usually detects wheels on trucks, armored personnel carriers and mobile missile launchers, but not on tanks. When viewed from the side, tanks and mobile missile launchers usually show a protrusion near the top of the vehicle. When viewed from above, they rarely show a protrusion. Trucks and armored personnel carriers rarely show a protrusion regardless of viewing angle. Tanks are usually seen off the road. Trucks, armored personnel carriers and mobile …show more content…
(15 points)
b. Calculate the posterior probability of vehicle type for the following cases. Do your results make sense given the problem description? Explain. (3 points each)
i. Object seen off the road, viewed from the side, protrusion is seen on the image, no wheels seen. ii. Object seen on the road, reported to be large, viewed from above, no protrusion seen. iii. Object seen on the road, reported to be small, wheels seen, viewed from side, no protrusion seen.
c. According to Rubronian military doctrine, a tank group consists of four tanks. An APC group has three armored personnel carriers and a truck. A mixed group has a tank, two
APCs, and a truck. Analysts knowledgeable about Rubronian military deployments have concluded that mixed groups are less common than tank groups, which in turn are less common than APC groups. When tanks are part of a mixed group, they usually travel on
OR/STAT 719 CSI 775 – Midterm Exam Spring 2014
Page 2 of 2 the road, whereas when they are in tank groups they usually travel off the road. Extend the model of part a) to handle groups of vehicles. Describe the extensions you made to your model – the nodes, the states and the probability distributions. (15