Assessment and Design of Color Vision Deficiency Simulation and Daltonization Methods
Doctoral thesis
Permanent lenke
http://hdl.handle.net/11250/2440785Utgivelsesdato
2017Metadata
Vis full innførselSamlinger
Sammendrag
Color vision in humans evolved because it improves behavioral responses related, for example, to the guidance of attention, object recognition and detection of emotional states. However, people with color vision deficiencies (CVDs) have a decreased ability in detecting chromatic edges and contrast. Thus, color-deficient people experience a certain loss of quality in images, confront difficulties in their day-to-day life and might also have some reduced behavioral performances. Methods to enhance images for color-deficient people – so-called daltonization methods – have been widely discussed, and color deficiency simulations for digital images have been introduced that help to understand the problems color-deficient people are facing.
One goal of this dissertation is the assessment of CVD simulation and daltonization methods regarding their impact on behavioral performances of the human visual system of color-deficient people. Moreover, we aim at developing daltonization methods with the focus of enhancing chromatic edges and contrast for people with different types and degrees of CVDs.
Firstly, we propose two behavioral methods for the assessment of CVD simulation and daltonization methods by using actual color-deficient people. We introduce a match-to-sample experiment that measures the indirect effect of simulation methods on the short-term memory, and a visual-search experiment that assesses the direct effect of daltonization methods on the attentional system. We show how the accuracy data from both experiments can be used as an interval scale to compare and rank different methods. At the same time, we point out the limitations of the response time data from both experiments.
Secondly, we propose two daltonization methods aiming at the enhancement of chromatic edges and contrast information in digital images. The first method, Yoshi-I, integrates chromatic edges and contrast into the lightness channel, which changes mostly the lightness of confusion colors. However, Yoshi-I did not always give satisfactory results due to visual clutter and unnatural colors. The second method, Yoshi-II, rotates and scales in the color space the error between the original and its simulation in the gradient domain towards the direction of optimal visibility, thus, changing confusion colors in hue while maintaining their overall lightness. Moreover, we offer an interface for data attachment that supports naturalness of memory colors like neutral colors. Yoshi-II returns robust and stable daltonization results, which is supported by behavioral and psychometric evaluation with color-deficient people