关键词:景观视觉质量;眼动实验;旅游景观;大学生;宏村 Abstract Landscape visual quality assessment is the basis of rational planning,development,utilization and protection of tourism resources. Various assessment methods have their own advantages,but many deficiencies. Most existing evaluation methods are subjectively qualitative approaches. Here,we introduce eye-tracking analysis to landscape visual quality assessment using the rural tourism destination Hongcun as a case study. An eyetracker was used to record eye movement data while subjects looked at Hongcun landscape pictures. We combined SBE and SD to obtain the subjective evaluation of landscape visual quality,which were analyzed by correlation and regression analysis with eye movement index data. We found that subjects had different eye movement characteristics in four landscapes types,among which,there were significant differences in fixation frequency,average fixation duration and saccade frequency. The fixation points mainly focused on the main buildings,some prominent elements and the borders of landscape elements. The subjective evaluation of landscape visual quality was significantly correlated with fixation frequency,percentage of fixation duration, average fixation duration and percentage of saccade duration. The eye movement evaluation model of landscape visual quality was built. Visual quality values of four landscape types in Hongcun, settlement, ecology, agriculture and culture, calculated by the model, were 5.99, 5.98, 5.96 and 5.90. The model shows that a higher fixation frequency and a longer watching time(a lower percentage of fixation duration) produce a higher landscape visual quality assessment of tourism landscapes. Our method is a new tool to conduct landscape visual quality assessments.
3.1.1 景观眼动指标差异 将四类景观的眼动数据进行单因素方差分析,得到表3。四类景观在注视频率、平均注视时间和眼跳频率这3个眼动指标上呈现显著差异(sig.分别为0.000、0.034和0.003),而在另外4个指标上差异不明显。不同景观之间差异程度不同,文化景观与其他景观相比眼动特征差异最大,其中注视频率和眼跳频率与另外三类景观均显著差异,文化景观的注视频率最低,平均注视时间最高,说明文化景观的信息量较大,被试需要花较多时间获取信息,眼跳频率最小表明文化景观的特征很明显,因此文化景观在宏村的景区设计中具有点睛之笔的功能,能增加景区的内涵和可读性,提高游客停留时间;农业景观平均眼跳幅度和平均眼跳速度均是最小,表明农业景观较为单调平和;聚落景观的注视频率较高但平均注视时间最短,表示聚落景观包含的信息量少,容易被观察;生态景观注视频率最高,表明被试对生态景观比较有兴趣,生态景观吸引力大。 Table 3 表3 表3不同景观类型眼动指标差异分析 Table 3Differential analysis of eye movement indexes in landscape types
眼动指标
FF
PFD
AFD
SF
PSD
ASA
ASV
总体差异(sig.)
0.000*
0.981
0.034*
0.003*
0.470
0.111
0.747
景观类型间差异
聚落景观(a)
2.54d
0.84
355.91
2.44d
0.07
2.65
84.00
生态景观(b)
2.57d
0.85
360.86
2.43d
0.07
2.86
86.47
农业景观(c)
2.47d
0.85
406.60
2.30d
0.06
2.41
82.24
文化景观(d)
2.21abc
0.85
424.48
2.08abc
0.07
3.97
88.48
注:*表示在0.05水上具有显著差异;文化景观中2.21abc,2.21表示文化景观的注视频率均值,abc表示文化景观与聚落景观、生态景观、农业景观有显著差异,其他值解释方式与此相同。 新窗口打开 3.1.2 景观眼动热点图 对总体被试的注视点位置和时间进行可视化得到景观的热点图(图1)。可以看出,聚落景观的视觉范围比较集中,注视点主要集中在月沼周边的古建筑和景观小品(小桥)上。生态景观和农业景观的注视点均未集中在某一特定要素上,而是看向田野、山体或水体交界的地方,其中生态景观注视点分布范围较大,视觉范围较为广泛。文化景观的视觉范围最为集中,注视点集中在石雕、牌匾区域上,完全忽略其他文化景观要素。因为受景观元素视觉特性影响,不同景观类型因景观元素的面积、形状、位置及色彩等不同而产生不同视觉感知和偏爱[49],聚落和文化景观的主题景观元素如房屋、雕刻等一般位于景观图片的中心,形状整齐,易形成视觉焦点,而生态和农业景观的主题景观元素如树木、山体、河流、田野等一般形状不规则,分布随意,且颜色相似,注视较为分散。总体来说,人们更偏向于观看景观的主体建筑物(图1a)、某种较为突出的景观元素(图1b、图1g、图1h)或景观元素交界处(图1c、图1d、图1e、图1f)。 显示原图|下载原图ZIP|生成PPT 图1四类景观的热点图 -->Figure 1Heat maps of four kinds of landscape pictures -->
3.2 眼动指标下的景观视觉质量评价模型
3.2.1 模型构建 根据主观评价下景观视觉质量评价体系公式(1)计算出每张图片的景观视觉质量值,将7个眼动指标的数值与景观视觉质量值进行相关分析,得到表4,发现主观评价的景观视觉质量与注视时间比重在0.01的水平上显著相关,与注视频率、平均注视时间及眼跳时间比重在0.05水平上显著相关,说明眼动指标与主观评价所得到的景观视觉质量值具有一定相关性。 Table 4 表4 表4景观视觉质量主观评价值与眼动指标相关性分析 Table 4Correlation analysis between eye movement indexes and subjective evaluation value of landscape visual quality
FF
PFD
AFD
SF
PSD
ASA
ASV
VQ
Pearson相关性
0.105*
-0.127**
-0.113*
0.075
0.101*
0.033
0.072
显著性
0.023
0.006
0.014
0.102
0.028
0.469
0.119
N
472
472
472
472
472
472
472
注:**、*分别表示在0.01、0.05水平(双侧)上显著相关。 新窗口打开 (1)假设眼动指标下的景观视觉质量(Eye-traking Visual Quality,EVQ)与7个眼动指标具有多元线性相关关系。 (2)将主观评价得到的景观视觉质量作为因变量,7个眼动指标作为自变量,利用逐步回归方法对自变量和因变量进行多元线性回归分析,希望得到景观视觉质量和眼动数据之间关系。 (3)在方差分析中,统计量F=7.465,相伴概率值sig.<0.005,进一步判断眼动指标与旅游景观视觉质量之间存在回归关系。从回归分析得到的表5可以看出,多元线性回归方程经过两个步骤得到,多元线性最优回归方程通过表5中的模型2可以给出,即景观视觉质量的眼动评价模型: (2) 式中 为眼动指标下的景观视觉质量;FF为注视频率;PFD为注视时间比重。 Table 5 表5 表5眼动评价模型回归系数分析 Table 5Eye movement evaluation model coefficients by regression analysis
(1)本研究以旅游景观为例,应用第一手实测数据,尝试将眼动评价法用于景观视觉质量评价。研究不仅为旅游景观视觉质量评价提供了客观、科学的方法,更为景观视觉质量评价提供新思路。根据心理学原理,凡能满足个体需要而使其产生积极愉悦情绪等情感体验的事物都会使其产生兴趣,吸引其注意而给予更多关注,同时表现出有一定规律的眼动模式[33],广告和网页的眼动研究中也证实了这一点[22,23,50]。本研究通过眼动探讨被试对景观的兴趣特征和景观的吸引力,进而评价景观的视觉质量,证明眼动数据对景观视觉质量具有一定指示作用,眼动分析法对景观视觉质量评价具有适用性。 (2)本研究选择了大学生为实验对象,虽然大学生作为被试进行眼动研究的尝试性探讨具有一定的代表性也非常有意义,但是大学生被试仅仅作为初步探讨眼动用于景观评价的可行性的一个案例。因为不同人口特征的群体对景观关注度是有差异的,仅选择大学生这一群体被试,其观点不一定能完全代表整体旅游者,初步的研究可能具有一定的局限性,作为景观眼动研究的尝试性探讨仍需进一步深化和拓展。今后相关的研究可以扩大被试人口背景范围,在真实景区以游客为被试进行实验,尽可能得到旅游者对景观更真实的评判,期望未来可以得出更具普适性的研究结论。 (3)研究实验图片不能完全代替实际景观、不同类型旅游地景观差异等问题仍待考虑,眼动用于景观评价尚有较大的研究空间。 The authors have declared that no competing interests exist.
Itami RM.Scenic Perception:Research and Application in U. S. Visual Management Systems[A]. Dearden P. Canadian Western Geographic Series [C]. ,1986. [本文引用: 1]
NordhH,Hagerhall CM,HolmqvistK.Tracking restorative components:Patterns in eye movements as a consequence of a restorative rating task [J]. ,2013,38(1):101-116. [本文引用: 1]
[31]
DupontL,AntropM,Van EetveldeV.Eye-tracking analysis in landscape perception research:Influence of photograph pro-perties and landscape characteristics [J]. ,2014,39(4):417-432. [本文引用: 1]
[5]
[Wu BH,Li MM.EDVAET:A linear landscape evaluation technique-a case study on the Xiao-xinganling scenery drive [J]. ,2001,56(2):214-221.] [本文引用: 1]
[6]
Daniel TC,ViningJ.Methodological Issues in the Assessment of Landscape Quality[A]. Altman I,Wohlwill J F. Behavior and the Natural Environment [M]. ,1983.
[32]
DupontL,AntropM,Van EetveldeV.Does landscape related expertise influence the visual perception of landscape photo-graphs? Implications for participatory landscape planning and management [J]. ,2015,141:68-77. [本文引用: 2]
[Li XQ,Zhao NX,Wang CZ,et al.A preli-minary study on the application of eye tracker in the campus landscape-taking the North Building of Nanjing University as an example [J]. ,2011,23(6):148-151.] [本文引用: 1]
[WangM.Research on the Eye Analysis Techniques Used in the Visual Quality Evaluation of Landscape-A Case Study of the Landscape of Danxia Landform,Sunan,Gansu[D]. ,2011.] [本文引用: 2]
[Feng JZ,Liu BY.Rationalization:A research on inventory methods of scenic resources [J]. ,1991,(5):38-43.] [本文引用: 1]
[12]
Daniel TC,Boster RS.[M]. New York:Plenum Press,1976. [本文引用: 1]
[37]
[ZhangJ,Lu SJ,Jiang ZJ,et al.The dimensions and characteristics of the perception of Chinese calligraphic landscape [J]. ,2012,67(2):230-238.] [本文引用: 1]
[38]
Ren XX,KangJ.Interactions between landscape elements and tranquility evaluation based on eye tracking experiments [J]. ,2015,138(5):3019-3022. [本文引用: 1]
[13]
Buhyoff GJ,Leuschner WA,Arndt LK.Replication of a scenic preference function [J]. ,1980,26(2):227-230. [本文引用: 1]
[14]
Osgood CE.Semantic differential technique in the comparative study of cultures [J]. ,1964,66(3):171-200. [本文引用: 1]
[Li ZP,Liu LM,Xie HL.Methodology of rural landscape classification:A case study in Baijiatuan Village,Haidian District,Beijing [J]. ,2005,27(2):167-173.] [本文引用: 1]
[Wang JY,LinL,GaoH,et al.Differences in college students’ spatial symbol cognition of tourism map:Based on experimental data from an eye-movement tracking system [J]. ,2016,31(3):97-105.] [本文引用: 1]