Abstract: Human lived in a variety of real situations, and they were influenced by all kinds of information from real-world scenes. However, the real-world scene included not only physical information, such as color, luminance etc., but also semantic information closely related to individual knowledge. So people's recognition for real-world scenes was the result of the interaction between top-down processing and bottom-up processing. The purpose of the present study was to explore the influence of scene physical features and the semantic properties on scene recognition. Two experiments were conducted to explore how scene information processing influenced recognition for real-world scene. In experiment 1 compared the recognition RT of the scene images with semantic relation or not under different rendering time conditions, and analyzed the effect of local luminance change on the scene recognition. The results showed that the recognition RT of the scene images with semantic connection was faster than the RT of the images without semantic connection, F(1, 87) = 328.82, p < .05, η2 = .23; the main effect of rendering time was found, F(1, 87) = 7.58, p < .05, η2 = .07, and the RT under 120ms rendering time condition was faster than under 60ms rendering time condition. Analysis of correct rate and RT of the measure pictures showed that the recognition rate of the mosaic pattern was higher, F(2, 29) = 80.44, p < .05, η2 = .105, and the reaction time was faster, F(2, 29) = 90.99, p < .05, η2 = .106, but the recognition rate of inverted image was lower and the reaction time was slower. In experiment 2, the researchers tried to explore the effects of local luminance change and rendering time on obtaining of scene background cues or semantic relation. The results showed that the rendering time(F(1, 87) = 6.90, p < .05， η2 = .01) or local luminance change (F(2, 87) = 5.04, p < .05，η2 = .02) has the main effect on the scene semantic getting, and the interaction between the rendering time and local luminance change was not significant, F(2, 87) = .33, p ＞ .05, η2 = .001. Subjects could obtain harder scene semantic relation when scenes were presented quickly, or the luminance of scene was 60%. On the other hand, the background cues obtaining time was influenced by the main effect of local luminance change (F(2, 87) = 26.59, p < .05, η2 = .10), and the interaction of two different factors(F(2, 87) = 10.79, p < .05, η2 = .05). Scene background cues were obtained under 40% luminance conditions more easy than under 60%, 80% luminance conditions. The above-mentioned results indicated that scene recognition influenced by scene semantic relation, and the effect of local luminance change on scene recognition was inhibited by scene semantic getting. During the scene recognition process, the obtaining of background cues or scene semantic relation was influenced by the rendering time and local luminance change. The slower the render time, the smaller the local luminance change, the background information or scene semantic relation was easier to get.

Key words: local luminance, semantic relation, scene recognition

摘要： 以真实情景图片为材料，探讨局部亮度与语义关系对情景识别的影响，以及情景背景线索和语义关系的获得。实验一结果显示，被试对有无语义关系情景图片的识别反应时存在显著差异，且对拼接图片的识别正确率更高，反应时更短。实验二探讨了局部亮度与呈现时间变化条件下情景背景线索和语义关系的获得。结果发现，情景语义关系和背景信息的获得受到局部亮度变化和呈现时间的影响。这表明，语义关系和呈现时间对情景识别存在显著影响，而且这种影响与情景背景线索和语义关系的获得有关。呈现时间越长，越易获得情景语义关系。局部亮度变化越小，背景线索越易获得，呈现时间的影响越小。

关键词: 局部亮度, 语义关系, 情景识别