ANALYSES OF 3-D OBJECT RECOGNITION PROCESS BY PYSCHOPHYSHICAL EXPERIMENTS

心理物理实验揭示了人类物体识别的视觉依赖和视觉独立特性，并指出这些研究在物质对象和实验范式上存在根本差异。因此，有可能在这两类实验中研究了不同的目标识别过程。为了评估这种可能性，两个研究了熟悉物体命名和新物体识别的实验结果表明，人类物体原型的描述包含了丰富的组件结构信息，但仅在有限的范围内发现了与视图无关的属性。因此，在这些原型被激活后，通过心理旋转的验证过程被认为是必要的，以检查与观察视图的对应关系。提出了原型激活和验证两个过程，可以调和实验之间的矛盾。它们显示了视图依赖和视图独立。指出三维物体识别的反应时间依赖于视点。我们认为，依赖性反映了将观察到的观点转化为规范观点的心理旋转过程的参与，并研究了三维物体被心理旋转的可能途径。我们的实验结果表明，命名熟悉的旋转物体的反应时间可以通过假设以下基于观察者为中心的坐标轴的心理旋转的模型来预测：首先(a)绕矢状轴旋转，然后(b)绕水平轴旋转，然后(c)必要时绕垂直轴旋转。此外，我们还分析了三维物体优度评级的决定因素。“一般视点率”是通过计算在一个视点上最具遗传性的视点的概率来定义的，该视点率描述了物体的所有潜在视点。研究发现，遗传视点率可以很好地预测三维物体的优等等级。最后，对婴儿的物体识别过程进行了研究，发现低频分量在物体统一性沉淀中起着非常重要的作用。少

Psychopysical experiments which revealed view-dependent and view-independent properties in human object recognition, and it is pointed out that fundamental differences can be found between these studies concerning the material objects and the experimental paradigms used. Therefore, there is a possibility that different process of object recognition have been investigated in these two types of experiments. In order to evaluate this possibility, two experiments which investigated naming of familiar objects and recognition of novel objects The results suggested that human prototypes of object have descriptions that contain rich information on component structure but view-independent properties ate found only in limited ranges. Therefore, after these prototypes are activated, a verification process by mental rotation is considered necessary to check correspondence to observed views. The proposed two process, the prototype activation and the verification processes, can reconcile contradicto … More ry points between the experiments. Which show view-dependence and view-independence.It has been pointed out that the reaction times for recognizing 3-D objects depend on view points. We have considered that the dependency reflects involvement of the mental rotation process which transforms an observed view into a canonical view and investigated possible pathways in which a 3-D objects are mentally rotated The results of our experiment s suggest that the reaction time for naming rotated familiar objects can be best predicted by the model which assumed the following mental rotations based on the observer-centered coordinate axes : first (a) rotation around the sagittal axis, and next (b) around the horizontal axis, then (c) around vertical axis, if necessary.Furthermore, we have analyzed the determinants for goodness rating on 3-D objects. 'Generic-view rate' was defined by calculating a probability of the most genetic view on a viewing sphere which describes all potential views of an object It was found that the goodness rating of 3-D objects can be well predicted by the genetic view rate.Finally, infants'' object recognition process was investigated and found that low-special frequency components take an very important role for object unity precipitin. Less