Measurement of variations in free-hand renderings using image analysis of geometric tasks
Lee R. Clendenning · Journal of computing sciences in colleges · 2005
The ability to analyze pixel image data for variations from the expected or ideal pixel locations is useful in such diverse disciplines as machine vision, industrial quality control, military target selection, child psychomotor development, geometric instruction, and computer games. This tutorial is a gentle, but interesting, exercise introducing students to the elementary concepts of bit-mapped formats, using pixel level data, and establishing local standards for valid statistical measures. Beyond presenting one of the simplest black and white bit-mapped formats, the tutorial emphasizes development of a locally standardized way to capture each image, use of a mathematical model to describe the ideal pixel locations, a sampling scheme, a nested loop scanning scheme to locate pixels of interest, and an operational definition of variation. The looping scans and measurement algorithms are implemented in C++, but could be implemented in other languages. The example images, on 5X 8 index cards, are responses from such free-hand tasks as drawing a line between two given points, or drawing a line parallel to a given line through a given point. Average errors smaller than .002 inches and standard deviations near zero have been achieved. Once students understand the basic concepts, they can easily progress to more advanced statistical and practical image analysis applications.