Searching for salient objects on cartographic backgrounds used in online map services: An experimental study based on eye movement analysis

Abstract

Our current understanding of the perceptual and cognitive processes involved in map reading remains incomplete. A map serves as a symbolic depiction of various aspects of the physical world and its corresponding characteristics. Despite its abstract nature, a map can still be a visually complex stimulus. When investigating map-related phenomena, experimental research draws upon established techniques from psychology and neuroscience, fields dedicated to studying human behavior and the workings of the human brain. The current empirical study focuses on examining and modeling aspects of visual perception related to the cartographic backgrounds utilized in popular online services. More specifically, the research involves designing and conducting a cartographic experiment that incorporates the use of eye tracking technology for recording and analyzing eye movements. The experiment's main goal is to identify characteristic values for key indices used in eye movement analysis, as well as to highlight salient areas of interest of the examined visual stimuli (maps). These indices are derived from two fundamental types of eye movements: fixations and saccades. Furthermore, the data collected in the experiment is utilized to generate cumulative grayscale statistical heatmaps, which provide an overall visualization of attentional behavior patterns, simultaneously highlighting the areas of interest of visual attention. The cartographic experiment engaged a group of 30 participants who were asked to freely observe 75 visual scenes featuring cartographic backgrounds sourced from five popular online services: Google Maps, Wikimedia, Bing Maps, ESRI, and OpenStreetMap. The visual scenes encompassed map backgrounds representing three distinct zoom levels: 12z, 14z, and 16z. Additionally, for each zoom level, five different Greek cities were selected, resulting in a total of 15 unique cities included in the visual scenes. It's important to note that the participants were not given any specific tasks and were simply instructed to observe the scenes. The objective of determining standard indices for describing the visual perception on cartographic backgrounds, along with the application of techniques to assess overall observational behaviors, enables comparisons to be made between different visual scenes. Analyzing these indices and observational behaviors, insights can be gained regarding the unique characteristics and challenges associated with the map reading process. Furthermore, modeling visual perception using data obtained from experimental studies and employing objective methodologies can provide valuable inputs for artificial intelligence systems. These inputs aim to predict the behavioral responses of map readers. This aspect is crucial for two main purposes: Firstly, for the objective evaluation of different options for cartographic representation, allowing for informed comparisons and secondly, for the development of user-oriented maps tailored to specific needs and preferences. The indices calculated present a uniform result, indicating that the visual exploration requires the same amount of attention and cognitive load for each visual stimulus. However, grayscale statistical heatmaps have proven that the behavioral patterns tend to differ, not only for every zoom level, but also for each distinct online map service too. Comparing the prominent features of the same geographical area for each service, salient objects differentiate in each visual stimulus.