Distance sets for shape filters and shape recognition
read more
Citations
Shape Classification Using the Inner-Distance
Vision-Based Traffic Sign Detection and Analysis for Intelligent Driver Assistance Systems: Perspectives and Survey
An Efficient Earth Mover's Distance Algorithm for Robust Histogram Comparison
Gromov–Wasserstein Distances and the Metric Approach to Object Matching
Contour detection based on nonclassical receptive field inhibition
References
Receptive fields, binocular interaction and functional architecture in the cat's visual cortex
Image Analysis and Mathematical Morphology
Combinatorial optimization: algorithms and complexity
Shape matching and object recognition using shape contexts
Receptive fields and functional architecture of monkey striate cortex
Related Papers (5)
Frequently Asked Questions (11)
Q2. What are the common data structures used to efficiently search nearest neighbors in higher dimensional spaces?
Advanced data structures which can be used to efficiently search nearest neighbors in higher dimensional spaces includequad-trees [43], K-d trees [44], range trees [45], lookup maps [46], etc.
Q3. How was the gradient of the oriented edge maps determined?
In order to eliminate small amplitude edges, a threshold condition was imposed on the gradient of the oriented edge maps (gradient be at least 15% of the maximum gradient value of the corresponding edge map).
Q4. How long does it take to compute the dissimilarity between two objects?
The time needed for computing the dissimilarity between two objects, each being described by approximately 250 points and having associated distance sets of 100 distances, is about 0.7 s on a regular Pentium III/667 MHz computer.
Q5. What is the proposed shape comparison method?
Formulated as a minimum cost assignment in an associated bipartite graph, the proposed shape comparison method delivers a reliable shape dissimilarity measure which can be used for shape classification.
Q6. How can a set of distance sets be scale invariant?
A set of distance sets associated with a given object can be made scale invariant by, for instance, dividing all distances in the set by the distance between the two feature points that are furthest apart, the diameter of the feature point set.
Q7. What is the reason for the coarseness of the shape context?
With respect to radial (distance) information, the shape context may be regarded as a simplified, coarser version of the distance set descriptor: the coarseness is due to the histogram binning process.
Q8. What is the value of the subset of points that remain after one application of the shape filter?
remain after one application of a shape filter associated with the letter “t.” (c) The subset of points S = fq 2 S j 9p 2 S ;D (p; q) = 0g which remain after two applications of the filter.
Q9. What is the subset of points left after applying the labeled distance set filter?
The subset of points S = LF (S ) remaining after applying only 2 filtering steps with a filter associated with the labeled distance set LS of a handwritten character t; two types of labels, l —contour point and l —junction point were used in this case.
Q10. How does the proposed shape comparison method perform?
by introducing an appropriate dissimilarity measure between two sets of (labeled) distance sets, the authors proposed a new shape comparison method.
Q11. How many distances were built for each point of the combined edge map?
The authors assigned a different label, , to each of the four orientations, and a labeled distance set was built for each point of the combined edge map by computing the distances to the first nearest neighbor points occurring in each oriented edge map, .