Spatio‐temporal analysis of eye movement patterns based on the maximum entropy method and the higher moment features
Spatio‐temporal analysis of eye movement patterns based on the maximum entropy method and the higher moment features
复制标题
基于最大熵法和高阶矩特征的眼动模式时空分析
DOI：
10.1002/ecjc.4430770301
复制
发表时间：
1994
期刊：
影响因子：
--
通讯作者：
F. Kishino
中科院分区：
文献类型：
--
作者：
Hidetomo Sakaino;N. Tetsutani;F. Kishino
本文在观察眼标图案时考虑了眼动图案。提出了基于最大熵的时空谱分析方法以及结合高阶矩特征的统计分析方法，从分析的谱图中提取特征。相似性图案被用作眼睛标记图案，其中线段、多边形和圆形被用作基本图案。将眼标图案投影在视角为80度的大屏幕上。图案的密度在视角方面约为1.04度和0.35度。从时间和频率两个角度分析了运动的频谱模式。可以看出，作为结果，角度，形状和图案密度的数量更强烈地反映在注视和扫视运动的持续时间上，而不是它们在观察期内的频率。这一结果表明，图像特征将估计从运动模式。
This paper considers the eye movement pattern when observing the eye mark pattern. The spatio‐temporal method of spectrum analysis based on the maximum entropy is proposed as well as the statistical method of analysis combining the higher‐order moment features to extract the features from the analyzed spectrum pattern. The similarity pattern is used as the eye mark pattern, where segment, polygon, and circle are used as the primitive patterns. The eye mark pattern is projected on a large screen with the view angle of 80 deg. The density of the pattern is approximately 1.04 and 0.35 deg in terms of the view angle. The spectrum pattern of the movement is analyzed from the two standpoints of time and frequency. It is seen as a result that the number of angles, shape, and pattern density are reflected more strongly on the durations of the gaze and saccade movement rather than their frequencies in the observation period. This result suggests that the image features will be estimated from the movement pattern.