DocumentCode :
2208084
Title :
Parameter setting of self-quotient ε-filter using HOG feature distance
Author :
Matsumoto, Mitsuharu
Author_Institution :
Educ. & Res. Center for Frontier Sci., Univ. of Electro-Commun., Chofu, Japan
fYear :
2011
fDate :
11-15 April 2011
Firstpage :
127
Lastpage :
133
Abstract :
This paper describes parameter setting of self-quotient ε-filter (SQEF) using Histograms of Oriented Gradients (HOG) feature distance. Parameter setting problem is generally solved by maximization or minimization of some objective evaluation functions such as correlation and statistical independence. However, it is not always easy to set such objective evaluation functions when we handle feature extracted images like SQEF because it is difficult to evaluate whether the parameter is optimal or not. On the other hand, even when we cannot employ objective assumptions, we sometimes know that an image includes some subjective information. Based on the above prospects, we consider HOG feature vectors of self-quotient filter (SQF) and SQEF of human images, and propose feature distance based parameter setting to use the subjective information. Experimental results show that the proposed approach has a potential to handle the parameter setting of feature extraction filter.
Keywords :
feature extraction; filtering theory; image processing; nonlinear filters; feature extracted images; feature extraction filter; histograms of oriented gradients feature distance; nonlinear filter; objective evaluation function maximization; objective evaluation function minimization; parameter setting problem; self-quotient ε-filter; statistical independence; Feature extraction; Histograms; Humans; Noise; Optimized production technology; Pixel; Shape;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Computational Intelligence for Multimedia, Signal and Vision Processing (CIMSIVP), 2011 IEEE Symposium on
Conference_Location :
Paris
Print_ISBN :
978-1-4244-9913-7
Type :
conf
DOI :
10.1109/CIMSIVP.2011.5949243
Filename :
5949243
Link To Document :
بازگشت