The experimental CBIR system
is implemented in MATLAB, the components include:
1. Feature Calculation:
CBIR_featurecalc() --- feature calculation
main, calculates features for each image and save to .mat files
CBIR_colorhist()
--- 256-bin HSV color histogram calculation
CBIR_edgedirection() -- 4-bin edge histogram
CBIR_edgehist() -- mpeg-7 edge histogram
2.
Distance Measures:
CBIR_L2dist() --- L2 distance of a given feature vector pair
CBIR_cosinedist() --- Cosine distance of a given feature vector pair
CBIR_histintersection() --- intersection of two histograms
3. Query:
CBIR_Query(query_img_id,
distance_id, weight_value) --- image query main script, rank images based
on weighted distance of color and edge features
CBIR_Query_minmax(query_img_id,
distance_id) --- image query main script, find minimax distance
To try out the system:
- run script CBIR_Query() or CBIR_Query_minmax() from: ~xlx/course/vis/ee6850-hw3-xlx/matlab/
- or download code to run on your own machine.
2004-04-14: i'm getting complaints from course
instructors
that students use this code instead of writing their own for HW ... seems
this toy code isn't usually doing the common good or making some
unserious. taken down upon request.
2007-06: if you indeed need legacy toy code, drop me a line
stating the intended useage and i shall reply with the code
pack in a few days.
EE6850
Home | Back to Intro | Previous:
Summary