关键词: 人机交互, 视线追踪, 视线突显

Abstract: The information display technology has gradually developed in the human computer interface. In order to improve the user's operating performance, many traditional display techniques, including highlighting, focus-context technology and application adaptation, were studied by many researchers. However, those above techniques are unable to adapt themselves to changing needs of users when they search on a massive information display. Combined with the highlighting technique and eye tracking method, eye-controlled highlighting was used as a new resolution in information display interface. It highlights directly the information which users are searching, as shown in Figure 1 to 2. Compared with traditional highlighting, eye-controlled highlighting dynamically highlights information determined by users’ eye movement. Users search information more naturally and efficiently. This study consists of two experiments. It attempts to study how different material and word frequency influence three display conditions in visual search. Those tests were developed in VisualC#2010. All sessions were run on a 1440×900 resolution monitor. A 2 (stimulus type: number vs. word in Exp.1, high-frequency word vs. low-frequency word in Exp.2)*3 (highlighting type: block highlighting vs. single highlighting vs. no-highlighting) between-subject design was adopted in both experiments. Two experiments’ dependent variables were accuracy, searching time, and user’s evaluation of NASA-TXL. The NASA-TXL is six aspects of evaluation: mental demand, physical demand, temporal demand, performance, effort and frustration. One hundred and ninety-two right-handed adults participated in the study (42 females and 54 males in Exp.1, 39 females and 57 males in Exp.2), randomly assigned to the 6 levels of experiment. For each group, participants were asked to finish 20 times of searching tasks with different displays’ conditions. After the searching tasks, participants’ subjective evaluation was performed. All data were processed by SPSS 13.0. The mean and standard deviation of the six experiment types’ performance and satisfaction evaluation were shown in Table 1 to 4. The two - factor variance analysis results showed as follows. In Exp.1, 1) Searching numbers and words with two eye-controlled highlighting, participants had higher accuracy than those with no-highlighting (ps< .05). Besides, when searching numbers and words with block highlighting, participants was faster than those with single highlighting (ps< .05). 2) There were no significant difference between two eye-controlled highlighting both on the performance searching numbers and words (ps >.05). 3) Participants searching number felt tired with single highlighting than with no-highlighting in the term of physical demand, but feeling more comfortable in the term of frustration with single highlighting (ps<. 05). In Exp.2, 1) Searching high-frequency and low-frequency words with block highlighting, participants were faster than those with single highlighting and no-highlighting (ps< .05). 2) Searching low-frequency words with single highlighting, participants had higher accuracy than those with block highlighting and no-highlighting (ps< .05). 3) Participants felt more comfortable in aggregate score with two highlighting than those with no-highlighting (p<. 05). The study shows these conclusions: (1) two types of eye-controlled highlighting are better than no-highlighting; (2) block highlighting are faster than single highlighting; (3) single highlighting has more advantages in searching low-frequency words.

Key words: human-computer interaction, gaze tracking, eye-controlled highlighting technology