Suppr超能文献

初级视觉皮层中图像片段的分组

Grouping of image fragments in primary visual cortex.

作者信息

Sugita Y

机构信息

Laboratory for Neural Information Processing, National Institute of Bioscience and Human Technology, Tsukuba, Japan.

出版信息

Nature. 1999 Sep 16;401(6750):269-72. doi: 10.1038/45785.

Abstract

In the visual world, objects are partially occluded by nearer objects, separating them into image fragments. However, the image fragments of the object can easily be grouped and organized together by the visual system. Psychophysical data and theoretical analysis indicate that such perceptual grouping might be mediated in the early stages of visual processing. Here I show that some orientation-selective cells in the primary visual cortex (V1) have response properties that can mediate the grouping of image fragments. These cells stopped responding to a stimulus bar when it was partly occluded by a small patch. The cells also did not respond when the patch had uncrossed disparity so that it appeared to be behind the bar. However, the cells began responding again when the patch had crossed disparity so that it appeared to be in front of the bar. These results indicate that cells as early as V1 have the computational power to make inferences about the nature of partially invisible forms seen behind occluding structures.

摘要

在视觉世界中,物体常被较近的物体部分遮挡,从而被分割成图像片段。然而,视觉系统能够轻易地将物体的图像片段组合并组织在一起。心理物理学数据和理论分析表明,这种感知分组可能在视觉处理的早期阶段就已介导。在此,我表明初级视觉皮层(V1)中的一些方向选择性细胞具有能够介导图像片段分组的反应特性。当刺激条被一个小斑块部分遮挡时,这些细胞会停止对其作出反应。当斑块具有非交叉视差,使其看起来在刺激条后方时,这些细胞也不会作出反应。然而,当斑块具有交叉视差,使其看起来在刺激条前方时,这些细胞又会开始作出反应。这些结果表明,早在V1阶段的细胞就具备对在遮挡结构后看到的部分不可见形状的性质进行推理的计算能力。

文献AI研究员

20分钟写一篇综述,助力文献阅读效率提升50倍。

立即体验

用中文搜PubMed

大模型驱动的PubMed中文搜索引擎

马上搜索

文档翻译

学术文献翻译模型,支持多种主流文档格式。

立即体验