摘要
为考虑边缘视觉的影响,实现对人眼宽视场条件下视觉成像质量的量化,提出一种基于孪生神经网络的多视域成像质量评价方法。构建个性化眼模型,根据波前像差值获得不同视场处的成像图;利用色彩差异分割成像图中的不同区域,将其作为子图像以样本对的形式输入到孪生神经网络中,提取图像的多维特征;模拟人眼对色彩的差异化感知,对区域图像质量评价值进行加权,得到对整幅图像的质量评价。为验证算法的有效性,在TID2013、LIVE和CSIQ这3个图像数据库上进行实验,其结果表明,该方法对多视场处成像质量的量化评估有良好的性能。
To consider the influence of peripheral vision and realize the quantification of visual imaging quality under the condition of the wide field of view,a multi-view imaging quality evaluation method based on the siamese neural network was proposed.A personalized eye model was constructed,the imaging at different fields of view was obtained according to the wavefront aberration value.Different regions in the image were segmented by color difference and input into the siamese neural network as sub-images in the form of sample pairs to extract multidimensional features of the image.The quality evaluation of the whole image was obtained by weighting the value of regional image quality evaluation by simulating the human eye’s differentiated perception of color.To verify the effectiveness of the algorithm,experiments were carried out on three image databases of TID2013,LIVE,and CSIQ.Results show that the method has good performance for the quantitative evaluation of imaging quality in multiple fields of view.
作者
王杨
隆海燕
贾曦然
WANG Yang;LONG Hai-yan;JIA Xi-ran(College of Electronics and Information Engineering,Hebei University of Technology,Tianjin 300401,China;Tianjin Key Laboratory of Electronic Materials and Devices,Hebei University of Technology,Tianjin 300401,China)
出处
《计算机工程与设计》
北大核心
2024年第4期1157-1165,共9页
Computer Engineering and Design
基金
河北省教育厅重点基金项目(ZD2020304)。
关键词
孪生神经网络
图像质量评价
个性化眼模型
色彩差异
边缘视觉
波前像差值
差异化视场成像
siamese neural network
image quality evaluation
personalized eye model
color difference
peripheral vision
wavefront aberration value
differentiated field of view imaging
