2001 Conference Proceedings

Go to previous article 
Go to next article 
Return to 2001 Table of Contents


DETERMINING OPTIMAL DISPLAY MODIFICATIONS AND ASSESSING THEIR IMPACT

Ian L Bailey
Kuang-mon Tuan
Richard T Wacker
Scott N Fitz
Angelika Angermann
School of Optometry
University of California, Berkeley, CA 94720

Lawrence H Boyd
Pasadero Inc. ,Tempe, AZ

Christopher Piedmonte
Eagle Creek Systems Inc. , San Diego, CA

Summary

We describe the development and applications of ColorFitter, a program to select optimal display colors, and LabTester for assessing the functional and subjective impacts.

Given the rapidly increasing use of computer displays in our society, it is appropriate to give attention to special needs of reading disabled people to enhance their access to information from computer systems. Computer technology provides enormous new flexibility to tailor the various characteristics of screen displays in order to suit the needs or preferences of individual subjects.

Persons with visual or reading disabilities can benefit from customized modifications of their computer displays. Larger print and reversed contrast have long been employed to make reading tasks easier for persons with reduced visual acuity. Over the past few years several computers and software products have been developed to facilitate the changing of visual displays by changing print size, font selection, font styles, spacings, layout, scrolling methods, synthesized speech augmentation, colors for letters, backgrounds and surrounds. Such display parameters can be tailored to best suit the individual vision characteristics of visually impaired and reading disabled persons.

In the Assistive Devices Program of the Low Vision Clinic of the University of California Berkeley, we conduct evaluations of visual abilities and attempt to determine which visual display parameters are optimal. Our normal clinical measurements of visual acuity and contrast sensitivity help determine print size and contrast requirements. Color is a parameter that is relatively easy to manipulate on computer display screens and we have developed a program called ColorFitter that allows for the systematic evaluation of an individual's color needs. A second program called LabTester is used for quantitatively evaluating consequences of changing display parameters. The program administers tests of performance and obtains feedback from the subjects as they perform visual tasks with the various display parameters under consideration. These two programs have been developed by Pasadero Inc. and Eagle Creek Systems Inc. working with a group from the Low Vision Clinic at UC Berkeley.

We shall describe these two programs and their application to choosing display parameters for reading disabled persons.

RATIONALE FOR THE DEVELOPMENT OF COLORFITTER

Tinted lenses are often recommended for individuals with certain eye diseases. Sometimes the tints are neutral or gray but at other times there are special chromatic requirements. Orange or amber filters are often recommended in retinitis pigmentosa and other retinal diseases. In macular degeneration, yellow filters can be beneficial. Persons with achromatopsia typically have a pronounced aversion to light, and red or magenta filters give best functional vision and comfort. Reading disabilities associated with dyslexia are sometimes treated with tinted lenses. Tinted lenses and colored filter overlays have been advocated by Irlen. Solan and his colleagues have used blue filtered and find improved reading performance, better comprehension and altered eye movement behaviors in reading disabled children. Wilkins and his co-workers have a method for prescribing tinted lenses and overlays using an optical system for selecting a most favored color for individual subjects. They find the tinted lenses can give improved performance and comfort in dyslexics and in persons with migraine.

DEVELOPMENT OF COLORFITTER

Color Coordinate System

For this program we developed our own coordinate system for specifying color and we call this the CPB system and the initials indicate Color, Purity and Brightness. In the Color domain, we have

effectively distorted the Hue dimension of HSL space by applying a polynomial function to force equal spacing of Red, Yellow, Green and Blue. Effectively we create a color circle where Red = 0, Yellow = 90, Green = 180 and B = 270 degrees. This causes complementary colors to be opposite each other and we have white in the center. Along a radius from the circumference to the center, the purity varies as the hue remains the same and the appearance gets paler moving towards white. The third dimension is Brightness, and represents a darkening created by turning down the signals to all three guns (R,G and B) in proportion.

The Subjects' Task

Subjects are presented with printed material seen against display fields with different color characteristics and they assign a score of relative preference - or goodness of appearance. There is one of the alternative displays chosen to be the reference and it has a reference point-score of 50 points.There are 4 different routines available.

(1) Color Sweep
In sequence of 16 different Colors (C values), each is compared to a reference white field. The P and B values can be selected to make the color appearance less Pure or less Bright. The reference white color has an assigned value of 50 points

(2) Color Select
There are 5 different C values equally spaced from a region of the spectrum that the user defines. The Purity and Brightness levels are specified. The color in the center of the range becomes the reference (50 point) field. This enables fine or coarse tuning of color preference within the selected region of the spectrum.

(3) Purity Select
There are 5 different Purity values equally spaced from a region of the Purity scale that the user specifies. The Color and Brightness levels are specified. The Purity value in the center of the selected range becomes the reference (50 point) field. This enables fine or coarse tuning of Purity preference within a selected region of the Purity range.

(4) Brightness Select
There are 5 different Brightness values equally spaced from a region of the Brightness scale that the user specifies. The Color and Purity values are specified. The Brightness value in the center of the selected range becomes the reference (50 point) field. This enables fine or coarse tuning of Brightness preference within a selected region of the Brightness range.

Alternative Modes for Presenting the Comparison Fields

There are three basic modes of presentation available in ColorFitter.

Five-Segment display.

This allows simultaneous presentation of the reference (50 point) display as a rectangular block in the center of the screen and surrounded by four other comparison display fields, one in each

corner. A sequence of four such displays is required to complete the Color Sweep that presents 16 colored fields for evaluation. Only one display is needed for each of the Selection routines.

Bipartite Field Display.

The screen display may be split in two - either vertically of horizontally. On one half field will be the reference condition (50 points) On the other half is one of the colored field samples from the current routine. In the case of a Color Sweep, there is a sequence of 16 displays in which one half of the screen has a white background, and the other has one of the 16 colors at the P and B values that the operator has selected. For any of the C, P or B Selection routines, one half of the screen always presents the midpoint sample a with its assigned 50 points. In a series of 4 sequential displays, each of the other 4 color fields are presented up against the reference and the subject assigns a point score. Here the subject is assigning points making binary comparisons of simultaneously viewed half fields.

Full Field Displays in Successive Presentations.

Here the full field display has only one color filling its field. The first display to be presented is the reference display filled with its pre-assigned score of 50 points. The subject uses the keyboard to switch to the comparison field with its different color appearance. The subject may go back and forth between the reference field and the test field as many times as needed before assigning a score to the test field. This mode of display and comparison is the most time-consuming. For a Color Sweep routine there must be at least 32 changes of field where every second field is white and every other is one of the 16 C values from around the color circle. For the 3 different Select routines, at least 8 display fields are required. The first has the midpoint value for the variable parameter as the reference field and it has a pre-assigned score of 50 points. In turn, the 4 comparison fields are shown. Based on successive observations, points are assigned to the 4 comparison fields. When each routine has been completed, the scores assigned by the subject are recorded in a tabular format. The results and the details of the display parameters presented up to this point can be viewed before selecting the fixed parameters and/or the range for the next routine. Within these results tables, the display parameters are expressed in CPB terms.

DEVELOPMENT OF LABTESTER

A major question is whether there is improved performance of visual tasks under favored display conditions and whether there are any measurable improvements in visual comfort or subjective perceptions. To test this, flexible and sophisticated computer software is necessary. Our group has developed a program that we call LabTester for testing subjects' performance at screen-based tasks and testing of visual resolution on display screens on which the appearance or display characteristics have been modified. Within LabTester there are a series of questionnaires that can be used to solicit the subjects' opinions and subjective impressions as they perform tasks or view different kinds of displays.

Performance tests

Within LabTester, there are two different tasks that we use to assess visual performance as we change the screen display conditions. We have a Reading and Comprehension Task and a Letter Counting Task.

Resolution tests

We have generated displays within LabTester that have printed material arranged in progressively decreasing size. We have tests for Letter Chart Acuity and Word Reading Acuity.

Questionnaires

Questionnaire screens can be authored in LabTester to suit the experiment being conducted. For evaluation of the effect of display parameters we mainly use questionnaires that ask for responses in the form of magnitude estimations. For this the subjects use a visual analog scale in the form of a screen-based slider-bar. By positioning the slider, subjects indicate the severity of visual

symptoms or discomfort, or they can indicate the strength of their opinions about the noticeability or the relative acceptability of the display features under consideration.

Test Sequencing

In LabTester, a sequence of events can be scheduled for an experimental session. In setting up an experiment, instructions need to be developed to specify the test tasks, the display parameters, and the associated questionnaires. Within an experimental session there may a series of trials using selected tasks that are performed under chosen display conditions. Interspersed strategically with the sequence of trials, questionnaires can be administered to get feedback from the subjects.

The Performance Tasks

Text Reading Task: Using the LabTester program, the subject first views an alert screen, which can carry basic instructions about the test task to be performed. Then the subject brings up a passage on the screen by pressing the space bar. Once the passage has been read, a second press of the space bar removes the displayed text and the duration of exposure is recorded. Then, three questions appear on the screen in turn and each question offers three "multiple-choice" answers. The subject is asked to select the correct answer. These questions serve as a measure of comprehension. We have created collections of close to 200 passages, about half being chosen excerpts from stories in Reader's Digest Condensed Books. We generated question sets for most of these passages. In setting up an experimental session, a set of instructions is required to specify the reading task. The instructions specify the path to the locations of the text passages, and these are linked to files that contain the comprehension questions for the passages.The text reading task is reasonably representative of reading prose on a screen display. It is designed to provide measures of reading speed and add a second measure to represent comprehension or attention.

Letter-Counting Task:

We have a letter-counting task in which the subject is required to count the number of occurrences of an assigned letter within a screen display of "pseudo-words." The text display is generated from letters being randomly arranged into strings of letters of variable length. The minimum and maximum "word" length is chosen, as well as the number of lines of "pseudo-words" presented, the minimum and maximum number of occurrences of the "target letter," and how many non-target letters should always be presented before possible repetition of the "target letter." Typically, only capital letters are used. We have used as search letters B, K, N, M, R, and W in Times Roman font, as preliminary experiments had found that these letters were about equally difficult to count within this task. The target letter is identified on the "Alert" screen presented at the start of each trial. By a press of the space bar, the subject initiates the display of the test task. A second press of the spacebar removes the task when the counting is completed. The exposure duration becomes the measurement of the search speed performance. For each task, the letter count is recorded by the subject using the keyboard to enter the count into a window on the display after the test screen has been removed. Unlike reading text, cues from word shapes and syntax are of no help and there are few clues that enable the subject to anticipate where a designated search letter will appear and what the total count will be for a given task segment. While it is not like real tasks commonly encountered in the workplace, it does demand very careful visual attention.

Specifying the Display Parameters for the presentation of the test tasks

A second set of instructions can be used by the experimenter to design the appearance of the task material presented on the screen. The experimenter can vary display parameters in various combinations. Variable display parameters include the font size (in points), style (normal, bold or italic), and type (e.g., Times New Roman, Arial), margin widths, line and letter spacing, background color, text color order of presentation for distinct tasks and number of repetitions of the tasks. There is an option for including a "spotlight" function in designated trials. We can create moving highlights of selectable lengths, colors, and mode and speed of movement.

A number of "Feedback" questions can be presented within the set of trials, and these can include questions about the text passage read, questions about the subject's rating of the display, or questions about visual symptoms. For ratings of subjective opinions, we use visual analogue scales. On a horizontal line on the screen, the ends and the center are designated with short vertical hash marks. The hash marks carry labels to indicate magnitude or degree of severity or preference. The subject uses a computer mouse to move the slider. One question is asked at a time.

Some Applications of ColorFitter and LabTester Programs

During the evolution and development of the ColorFitter and LabTester programs, numerous informal and semi-formal pilot runs were done with normally sighted and reading disabled subjects. These programs have been applied to more formal studies. The programs are used within a clinical setting for application to specific patient needs in the School of Optometry's Assistive Devices Program within its Low Vision Clinic.

Two research applications of the LabTester and Color Fitter will be described. The first involves testing for the effects of colored display s on the reading performance of dyslexic and normal subjects. The other involves testing for the effects of an electronic place-keeping aid (moving highlight) on subjects with Macular Degeneration.

REFERENCES

Irlen, H. (1991) Reading by the Colors. New York: Avery Publishing Group Inc.

Solan HA, Ficarra AP, Brannan JR, Rucker F, (1998) Eye movement efficiency in normal and reading disabled children: effects of luminance and wavelength. J Amer Optom Assoc, 69: 455-64.

Wilkins AJ, Evans BJW, Brown J, et al. Double-masked placebo-controlled trial of precision spectral filters for children who use coloured overlays. Ophthal Physiol Opt, 1994; 14:365-70.


Go to previous article 
Go to next article 
Return to 2001 Table of Contents 
Return to Table of Proceedings


Reprinted with author(s) permission. Author(s) retain copyright.