Reconstructing Shape and Texture of 3D Object by Active Wearable Vision Sensor

为了在可穿戴计算机环境中建立人机交互，我们研究了通过主动可穿戴视觉传感器信息重建用户手中的3D物体的形状和纹理。得到的研究结果可以总结如下。使用主动可穿戴视觉传感器检测人的眨眼点。我们开发了一种可穿戴的视觉传感器，它由人的视线探测器和两个活动摄像头组成。首先，我们建立了一种校准传感器的方法，使其即使在真实情况下，也能准确地检测到人的眨眼点深度的变化。接下来，我们提出了一种检测人注视区域的方法。手持式3D物体的形状重建。提出了一种保留目标表面局部结构的距离图像配准方法。该方法引入了形状模式和对应的偏度，这两者都是从局部表面图像中提取的，更重要的是每个图像中的兴趣点。这两个特征使我们能够在保留物体表面局部结构的情况下重建三维物体的完整形状。复杂光照下三维物体的反射率估计。我们开发了一种从固定相机在未知和复杂照明下拍摄的物体图像序列中恢复运动朗伯物体反射特性的方法。我们的方法使我们能够利用重建的形状和反照率在任意光照条件下合成任意姿态的物体的真实图像。基于主动可穿戴视觉传感器的自我运动估计。我们提出了一种通过安装两个有源摄像机来增量估计自我运动的方法。我们的方法在自我运动过程中独立控制两个相机，使每个相机自动将其光学轴固定在自己的固定点上。在这种情况下，两个相机不必共享共同的视野，而在立体视觉中，它们必须这样做。少

To establish man-machine interactions in the wearable computer environments, we studied reconstructing the shape and texture of a 3D object that a user takes in his hands through the active wearable vision sensor information. Obtained research results can be summarized as follows.1.Detecting blink points of a person using active wearable vision sensor.We developed a wearable vision sensor that consists of the detector of person's viewing lines and two active cameras. First, we established a method for calibrating the sensor so that it can detect person's blink points accurately even in a real situation such that the depth of blink points changes. Next, we proposed a method for detecting a gazing region of a person.2.Shape reconstruction of a hand-held 3D object.We proposed a registration method for range images that preserves local structures of object surfaces. The method introduces shape patterns and a skewness of correspondences, both of which are extracted from the local surface ne … More arby a point of interest in each image. These two features enable us to reconstruct full shape of a 3D object that preserves local structures of object surfaces.3.Reflectance estimation of a 3D object under complex illumination.We developed a method for recovering reflectance properties of a moving Lambertian object from an image sequence of the object taken by a fixed camera under unknown and complex illumination. Our method enables us to synthesize realistic images of the object in arbitrary poses under arbitrary lighting conditions by using the reconstructed shape and albedo.4.Ego-motion estimation using active wearable vision sensor.We proposed a method for incrementally estimating ego motion by two mounted active cameras. Our method independently controls the two cameras during the ego motion so that each camera automatically fixates its optical axisto its own fixation point. Two cameras do not have to share the common field of view in this case whereas in stereo vision they have to do. Less