Scale-space representation of 3D models and topological matching

2003 ◽  
Author(s):  
Dmitriy Bespalov ◽  
Ali Shokoufandeh ◽  
William C. Regli ◽  
Wei Sun
Keyword(s):  
3D Models ◽  
Scale Space ◽  
Space Representation

Scale-Space Representation and Classification of 3D Models

2003 ◽  
Vol 3 (4) ◽  
pp. 315-324 ◽  
Author(s):  
Dmitriy Bespalov and ◽  
Ali Shokoufandeh ◽  
William C. Regli ◽  
Wei Sun
Keyword(s):  
Shape Matching ◽  
Distance Measure ◽  
Empirical Evaluation ◽  
3D Models ◽  
Scale Space ◽  
Space Representation ◽  
Point Distribution ◽  
Solid Models ◽  
Recent Developments

This paper presents a framework for shape matching and classification through scale-space decomposition of 3D models. The algorithm is based on recent developments in efficient hierarchical decomposition of a point distribution in metric space p,d using its spectral properties. Through spectral decomposition, we reduce the problem of matching to that of computing a mapping and distance measure between vertex-labeled rooted trees. We use a dynamic programming scheme to compute distances between trees corresponding to solid models. Empirical evaluation of the algorithm on an extensive set of 3D matching trials demonstrates both robustness and efficiency of the overall approach. Lastly, a technique for comparing shape matchers and classifiers is introduced and the scale-space method is compared with six other known shape matching algorithms.

scholarly journals A parameterless scale-space approach to find meaningful modes in histograms — Application to image and spectrum segmentation

2014 ◽  
Vol 12 (06) ◽  
pp. 1450044 ◽  
Author(s):  
Jérôme Gilles ◽  
Kathryn Heal
Keyword(s):  
Image Segmentation ◽  
Color Image ◽  
Scale Space ◽  
Space Representation ◽  
Grayscale Image ◽  
Local Minima ◽  
Clustering Problem ◽  
Space Curves ◽  
Space Approach

In this paper, we present an algorithm to automatically detect meaningful modes in a histogram. The proposed method is based on the behavior of local minima in a scale-space representation. We show that the detection of such meaningful modes is equivalent in a two classes clustering problem on the length of minima scale-space curves. The algorithm is easy to implement, fast and does not require any parameter. We present several results on histogram and spectrum segmentation, grayscale image segmentation and color image reduction.

scholarly journals Penerapan Speeded-Up Robust Feature pada Random Forest Untuk Klasifikasi Motif Songket Palembang

2020 ◽  
Vol 5 (3) ◽  
Author(s):  
Yohannes Yohannes ◽  
Siska Devella ◽  
Ade Hendri Pandrean
Keyword(s):  
Random Forest ◽  
Hessian Matrix ◽  
Scale Space ◽  
Haar Wavelet ◽  
Space Representation ◽  
Interest Point ◽  
Interest Points ◽  
Random Forest Classification ◽  
Forest Classification ◽  
Historical Heritage

Songket is a historical heritage in the city of Palembang. Where Songket has many different types and motifs. Besides having historical value, Palembang's original Songket has high quality and complexity in the manufacturing process. As known Palembang Songket has a lot of motives, one of the ways to recognize Palembang Songket is through its motives, so that research was conducted for the classification of Palembang Songket motifs. The method used to extract features is the Speeded-Up Robust Feature (SURF), while the classification method is Random Forest. The process of forming the SURF feature is divided into two stages, the first stage is Interest Point Detection, which consists of Integral Images, Hessian Matrix Based Interest Points, Scale Space Representation and Interest Point Localization, the second stage of Interest Point Description consists of Orientation Assignment and Descriptor Based on Sum Haar Wavelet Responses. The resulting feature is used for the Random Forest classification. This study used 345 images of Palembang Songket motifs, among others, Bunga Cina, Cantik Manis and Pulir. The images taken are based on 5 colors from each Palembang Songket motif. For the separation of data there are 300 images used as data train and 45 images for testing data. From the tests that have been done the results of the overall overall accuracy are 68.89%, per class accuracy 79.26%, precision 69.27, and recall 68.89%.

Sign in / Sign up

Export Citation Format

Share Document