基于ANFIS-LSSVM的计算颜色恒常性算法研究  

A study on the computational color constancy algorithm based on ANFIS-LSSVM

作　　者：王兴光 罗运辉 王庆[1] 陈业红[1] WANG Xingguang;LUO Yunhui;WANG Qing;CHEN Yehong(Faculty of Light Industry,Qilu University of Technology(Shandong Academy of Sciences),Jinan 250353,China)

机构地区：[1]齐鲁工业大学(山东省科学院)轻工学部,山东济南250353

出　　处：《齐鲁工业大学学报》2024年第2期62-72,共11页Journal of Qilu University of Technology

基　　金：齐鲁工业大学教育部制浆造纸重点实验室开放基金项目(ZR202001)。

摘　　要：计算颜色恒常性是指消除场景光源的影响从而再现物体真实颜色的能力。目前,深度神经网络的应用使颜色恒常性精度显著提高,但大多数深度学习算法训练时间长、计算复杂度高,且需要大量的训练样本。针对此问题,提出了一种结合自适应神经模糊推理系统(ANFIS)和最小二乘支持向量机(LSSVM)的简单有效的方法。该方法分为训练和预测两个阶段:在训练阶段,首先提取图像特征分别训练ANFIS、LSSVM两种初始光源估计模型,接着利用核函数变换将两种模型融合,然后利用预留训练样本进一步训练得到多元线性回归光源估计模型;在预测阶段,提取测试图像特征后,直接由训练所得模型预测得到该测试图像最终的场景光源颜色值。实验结果表明,与深度学习方法相比,本文所提方法计算复杂度较低,即使在小训练样本中也能有很好的光源估计性能。Computing color constancy refers to the ability to eliminate the influence of scene light sources and thus reproduce the true color of an object.Currently,significant improvements in color constancy accuracy have been achieved by using deep neural networks,but most deep learning algorithms have long training time,high computational complexity,and require a large number of training samples.To address this problem,this paper proposes a simple and effective method that combines adaptive neuro-fuzzy inference system(ANFIS)and least squares support vector machine(LSSVM).The method is divided into two phases:training and prediction.In the training phase,image features are first extracted to train two initial light source estimation models,ANFIS and LSSVM,respectively.Then the kernel function transform is used to fuse the two models,and the multivariate linear regression model of light source estimation is obtained by using the reserved training samples for further training.In the prediction phase,after extracting the features of the test image,the final color values of scene light source of the test image is predicted directly from the training data.The experimental results show that,compared with the deep learning method,the proposed method is of lower computational complexity and even with small training samples,and it also has good light source estimation performance.

关 键 词：计算颜色恒常性 光源估计 自适应神经模糊推理系统(ANFIS) 最小二乘支持向量机(LSSVM) 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]