Visual thresholds for detection, recognition and visual impact in landscape settings

Although the concept of visual threshold has long been used in psychophysics, its transplanting to visual landscape research is very new. Very few studies have attempted to look at visual thresholds in landscape settings, and none has ever produced any tangible results. This study was the first to have proposed a systematic organising structure for visual impact assessment procedures via three articulated visual thresholds - visual thresholds for detection, recognition and visual impact. Three sets of visual thresholds, which were defined as the points when 50% of the viewers accomplished a requested perception task, were obtained through controlled slide-viewing testings using computer simulated images with modified visual attributes. Three contrasts, four visual sizes, two object types and four baseline landscape settings were the defined independent variables. Detection, recognition and visual preferential rating were the dependant variables. These variables were controlled, manipulated and combined using photorealistic computer imaging techniques from which a total of 88 simulations were produced. Two groups of randomly selected university students (94 altogether), half of whom pre-informed, were used as the test subjects on 88 pairs of "before" and "after" images in a strictly controlled environment. From the tests results, general and specific visual thresholds values were derived and four detailed multivariate logistic regression models for the thresholds were also established. In general, detection, recognition and visual impact thresholds were found to be around 27, 65 and 85 square minutes. But more precise results were given by the logistic regression equations which achieved an average prediction success of over 90%. Many perceptual phenomena were also observed. For example, contrast weighted visual size (measured in square minutes of angle subtended by an object multiplied by visual contrast) was found to be the determinant variable for visual detection, recognition and visual impact assessment; detection and recognition were found to be linearly related to the log visual size of the object; it was many times easier to detect an object on a darker background than on a brighter background, and pre-knowledge definitely increased people's ability to recognise objects. The term maximum visibility was redefined as the detection threshold, and the concept of perception radius (visibility, recognisability and impact threshold) was initiated and used. The obtained thresholds, equations and perceptual findings could be readily applied to landscape management systems and enable automated mapping and zoning of the maximum perception radii. Perhaps it is the establishment of the three articulated visual thresholds concepts and their quantitative definitions in landscape settings that will challenge and redefine the validity and true usefulness of the sensitivity levels and the current visibility theories upon which many existing visual management systems are built