Upload
wilfred-grant
View
217
Download
0
Embed Size (px)
Citation preview
REU WEEK IVMalcolm Collins-Sibley
Mentor: Shervin Ardeshir
GOALS FROM LAST WEEK
• Understanding the occlusion handling code
• Making sure it is handling self-occlusions accurately
• Understanding the format of the output data in the line segments/horizon code
• Running the line segmentation code for all of the images in our dataset and saving all of the output variables in a structure
• Extracting the super pixels from images in the dataset and saving it in a structure
• Computing their pairwise similarities of the super pixels in terms of color and texture
COMPLETED WORK• Error and inaccuracy fixing with the building projection code
COMPLETED WORK
COMPLETED WORK
COMPLETED WORKSmall changes to the top view map
COMPLETED WORK
• Probability Mapping• Each image has
one map with each building section covered by a Gaussian filter
COMPLETED WORK
Binary map of where there is a high probability of the building being there
COMPLETED WORK
Multiple building binary maps
COMPLETED WORK
COMPLETED WORK
• Data Storage
Number four is empty because no buildings were detected
FUSION – PROPAGATION
• We will build a graph on the super-pixels
• Nodes = Super-pixels (Probability of segment I belonging to a building-f(intersection) )
• Edges = Similarity of the super-pixels in terms of color, texture, location, etc
COMPLETED WORK
COMPLETED WORK
THE NEXT STEP
• Tuning building projections in terms of height.
• Generating KML/KMZ files from google earth containing GPS locations of different buildings/roads
• Fusion between building projection and super-pixilation • First with binary mapping• Next with probability mapping
• Initial fusion results (Belief Propagation)
• Run that fusion on the data set