Browsing by Author "Fan, KW"

Now showing 1 - 2 of 2

An illumination estimation scheme for color constancy based on chromaticity histogram and neural network
(2005)
Lin, CT
;
Fan, KW
;
Cheng, WC
;
電控工程研究所
;
Institute of Electrical and Control Engineering
This paper proposes an algorithm for estimation the illumination of an image for the purpose of color constancy using chromaticity histogram. This method evaluates the color temperature of the light source by detecting the distribution of chromaticity histogram in an image. It has the advantages of high efficiency, good robustness, and no strict assumptions. We subsequently propose to use a multilayer perceptrons trained by Backpropagation algorithm to model the nonlinear functional relationship between the chromaticity histogram and coefficients of illuminant functions. The trained neural network can then be used to estimate the spectral power distribution of light source. Finally, we use the trained neural network to estimate spectral power distribution in a finite-dimensional linear model of surface reflectance for color constancy. For performance evaluation, two color-recovery experiments on synthetic images and nature images captured from a still digital camera are performed in this paper. All the results are compared to those of two existing popular algorithms (Max-RGB and Gray-World algorithms).
Fuzzy neural network design using support vector regression for function approximation with outliers
(2005)
Lin, CT
;
Liang, SF
;
Yeh, CM
;
Fan, KW
;
電控工程研究所
A fuzzy neural network based on support vector learning mechanism for function approximation is proposed in this paper. Support vector regression (SVR) is a novel method for tackling the problems of function approximation and regression estimation based on the statistical learning theory. SVR has been shown to have robust properties against noise. A novel support-vector-regression based fuzzy neural network (SVRFNN) by integrating SVR technology into FNN is developed. The SVRFNN combines the high accuracy and robustness of support vector regression (SVR) and the efficient human-like reasoning of FNN for function approximation. Experimental results show that the proposed SVFNN for function approximation can achieve good approximation performance with drastically reduced number of fuzzy kernel functions.