Knowledge Base

[ Discussions | TOC | Search | Post | Reply ]

Article 27

Category: Weekly Report
From: Kevin Berry
Date: 4/10/2001
Time: 5:32:35 PM
Remote Name:


Weekly Team Report #12 Team #6

• Action Items: 1.Finalize setup for measuring device, recover LED strip from deep water from recent heavy rain and decide how to keep this from happening again in the future 2.Continue coding

• Outstanding Problems: 1.During heavy rains the LED stick for nighttime detection keeps washing away. 2.Daytime and Nighttime algorithms

• Plans to Address Outstanding Problems: 1.The LED stick has now been mounted on the tree to keep it from moving but with the cost of being higher off the ground. Wait and see how it works for the next rainfall. 2.Daytime Algorithm: -Detect distinct objects. -Find out which ones look like lines -Find a pattern of lines close together. -To get better accuracy, not only measure the number of lines, but also the length of the lowest ones compared to the highest, since the low ones are partially hidden, they should be shorter, which gives a dimension which is wider to work with. -For uncertainty, also a fast and easy way to do it: I'll just invert the image and run through the algorithm again. This time it will be detecting the white lines if it was doing the black before, or vice versa. Then one can just compare the two results, as in water height and location of the marker. Nighttime Algorithm: -Since the LED’s in the image or very close together and would be difficult to distinguish individual LEDs, every other one was covered in black tape. This should work but brings the accuracy down to 4 inches between each LED. Daytime vs. Nighttime Detection: -one way is to take random points and determine the color. If enough points are black then it is a good assumption it is night. Another way is to look at the size of the .jpg file. Daytime images are usually at least 40 Kb and nighttime images are around 4 Kb. Images in between would be during dusk or dawn where both detection algorithms could be run and the one with most accurate result used. • Hours Worked for Each Team Member 1.This week Ashes Ganguly – 8 Alex Giouzenis – 8.5 Kevin Berry – 8 Ruigang Yang – 8 2.Totals Ashes Ganguly – 53 Alex Giouzenis – 53.5 Kevin Berry – 53.5 Ruigang Yang – 54

• Milestones 1.Accomplished a)Coding and Algorithm detection analysis 2.Approaching a)Continuing programming 3.Problems/Slips a) none at this time

Last changed: June 17, 2001