摘要
目前的人机交互界面视觉显著度评估方法准确性差、能耗高,提出基于人眼视觉的多媒体人机交互界面视觉显著度评估方法。采用自适应高斯滤波器对人机交互界面进行滤波,通过Guassian函数将界面图像梯度方向转换成水平与垂直两个方向的导数,确定界面梯度方向大小,据此得到校正后的交互界面图像。利用Drago对数算子对校正后的交互界面色调进行调整,根据交互界面亮度与场景亮度之间的映射关系,得到界面像素值的亮度值压缩度可见细节程度,获取色调和亮度调整后的交互界面图像。依据界面图像视觉显著度的共现不同理念,将人眼视觉显著度影响指标进行量化,得到交互界面空间位置函数。设置交互界面各个图像块内方向种类数量和边缘点数量,获取界面各子图像中某像素点的纹理复杂度函数。给出加权系数,将界面空间位置函数和像素点的纹理复杂度函数作为评估指标进行加权输出,给出交互界面视觉显著度评估结果。实验结果表明,上述方法平均评估能耗为59.4nJ/bit,评估准确率高。
At present,accuracy of assessment method of visual saliency of human-computer interaction interface is poor.Meanwhile,energy consumption is high.In order to overcome the disadvantages,we proposed a new assessment method based on human vision.Gradient direction of interface image was converted into derivatives in horizontal and vertical direction respectively via Guassian function to confirm interface size in gradient direction.Based on that,the corrected interaction interface image was obtained.Hue of the corrected interaction interface was adjusted via Drago logarithm operator.According to mapping relation of brightness between the interaction interface and scene,visible details degree of brightness value compression of interface pixel value was obtained and interaction interface image after adjusting hue and brightness was acquired.Influence index of saliency of human vision was quantified to obtain spatial position function of the interaction interface according to different co-occurrence of visual saliency of interface image.Quantity of direction species and peripheral points in each image block of the interaction interface was set up and texture complication function of some pixel in each sub-image of interface was acquired.Weighting coefficient was provided.The spatial position function and texture complication function were used as assessment index to carry out weighted output,and assessment results were provided.Simulation results show that energy consumption of the method is 59.4 nJ/bit.The assessment accuracy is higher.
作者
王素丽
WANG Su-li(Guangzhou College,South China University of Technology,Guangzhou Guangdong 510800,China)
出处
《计算机仿真》
北大核心
2020年第3期161-164,共4页
Computer Simulation
关键词
多媒体
交互界面
视觉显著度
评估
Multimedia
Interaction interface
Visual saliency
Assessment
