摘要
针对卷积神经网络(CNN)中间层特征维度高,含噪声较多的问题,提出一种CNN特征降维的方法,首先利用主成分分析(PCA)对CNN特征进行降维,在数据层面和人类感知层面证明了其有效性;然后将降维后的CNN特征作为区域特征向量,利用多水平超像素分割和随机森林回归构建了一个融合手工特征及降维CNN特征的显著性检测模型;最后选取了10个显著性检测传统模型进行对比,构建的融合模型性能优于仅使用传统手工特征的方法,降维后的CNN特征能够改进显著性模型的性能.
To deal with the high dimensions and noise of CNN features,a method to reduce dimensions of CNN features is proposed,principal component analysis(PCA)was first used on CNN features to reduce the dimensions,the effectiveness of PCA was verified in both data aspect and human subjective perception.Then a saliency model was constructed by using the multi-level superpixels segmentation and random forest to fuse PCA-CNN features and handcrafted features.Experiments demonstrated the proposed saliency model outperformed other traditional saliency methods,the PCA-CNN features can improve performance of saliency detection.
作者
方正
曹铁勇
郑云飞
杨吉斌
Fang Zheng;Cao Tieyong;Zheng Yunfei;Yang Jibin(Institute of Control and Command Engineering,Chinese People’S Liberation Army Engineering University,Nanjing 210001;Key Laboratory of Polarization Imaging Detection Technology,Chinese People’s Liberation Army Artillery Air Defense Academy,Hefei 230031)
出处
《计算机辅助设计与图形学学报》
EI
CSCD
北大核心
2019年第2期324-331,共8页
Journal of Computer-Aided Design & Computer Graphics
基金
国家自然科学基金(61471394)
江苏省优秀青年基金(BK20180080)
关键词
显著性检测
卷积神经网络特征
主成分分析
特征融合
随机森林
saliency detection
convolution neural network features
principal component analysis
feature fusion
random forest
