Mike Mertsock

Over the past week, I focused on finalizing my project plan and goals, reading more papers, and learning the Matlab foundations of the spatial processing techniques I will be using.

Reading Completed

• “R-Histogram: Quantitative Representation of Spatial Relations for Similarity-Based Image Retrieval”, Wang and Makedon, 11th ACM Conf. on Multimedia, 2003—introduction of the R-Histogram and experimental results

Image processing techniques learned

• Connected component analysis; finding region boundaries using boundaries and other functions
• Finding centroids of segmented regions
• Getting familiar with the fuzzy inference tooklit
• Created trapezoidal fuzzy sets for spatial relationship classification (such as left, right, etc.), as described in Keller and Wang 1995
• Implemented the centroid method of classifying spatial relationships between regions, as described in Keller and Wang. Tested
• Most of the work of the aggregation method for spatial relationships, including finding a histogram of angular differences (the classical histogram, not the R-histogram) and calculating a vector of fuzzy set memberships (one vector of membership values per fuzzy set). Using arithmetic mean as a simplified aggregation operator produced the expected results.

Next steps

1. Retrieve k-Nearest Neighbor book by Dasarathy from the library
2. Implement R-Histogram
3. Collect test data

This entry was posted on Tuesday, February 7th, 2006 at 5:11 pm and is filed under CV 20052. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.