Abstract: A study on sampling order for irregular geometric figure recognition was conducted. Ss were 80 university students in Zhejiang University, China, half male and half female, aged from 18 to 22, with normal color vision and normal or correct-to-normal vision. The stimuli were 20 irregular geometric figures. The independent variables were: a) visual task, figure recognition and figure matching; b)visual condition, good and not good; c)display type, rotated and not rotated .The visual performance used as dependent variables was measured by 2 criteria: a) response time, b) wrong index mark. The experiment consisted of 2 tasks, judging the target figures displayed on CRT and doing a questionnaire about the sampling order. Friedman test and Kendall’s Coefficient of Concordance test showed that the sampling order for irregular geometric figure recognition was generally stable, it was accorded with the degree of the difference among the sampled-features. The sampled- features with higher degree of such difference would arrange prior order to those with lower degree. The study suggested that stable sampling order, especially with global information traits prior to local information traits, is helpful in improving the performance of figure recognition

Key words: figure recognition, sampling strategy, sampling order