A 3-Dimensional Object Recognition Method Using Relationship of Distances and Angles in Feature Points

In recent years, human support robots have been receiving attention. Especially, object recognition task is important in case that people request the robots to transport and rearrange an object. We consider that there are four necessary properties to recognize in domestic environment as follows. (1) Robustness against occlusion. (2) Fast recognition. (3) Pose estimation with high accuracy. (4) Coping with erroneous correspondences. As conventional object recognition methods using 3-dimensional information, there are model-based recognition methods such as the SHOT and the Spin Image. The SHOT and the Spin Image do not satisfy all four properties for the robots. Therefore, to satisfy the four properties of recognition, we propose a 3-dimensional object recognition method by using relationship of distances and angles in feature points. As per our approach, the proposed method achieves to solve problems of conventional methods by using not only the feature points but also relationship between feature points. To achieve this purpose, firstly, the proposed method uses a curvature as a feature in a local region. Secondly, the proposed method uses points having high curvature as feature points. Finally, the proposed method generates a list by listing relationship of distances and angles between feature points and matches lists.

Análisis y Comparación del Descriptor Cone Curvature frente al Reconocimiento de Expresiones Faciales

Se presenta el resultado de analizar el comportamiento del descriptor de forma Cone Curvature (CC) en la tarea de reconocimiento de expresiones faciales en imágenes 3D. El descriptor CC es una representación del modelo 3D que se calcula a partir de un conjunto de ondas de modelado para cada vértice de una malla poligonal. Se empleó la base de datos de rostros 3D (BU-3DFE), la cual contiene imágenes con 6 expresiones faciales. Con el uso del descriptor CC, las expresiones fueron reconocidas en un porcentaje promedio del 76.67% con una red neuronal, y del 78.88% con un clasificador bayesiano. Al realizar una combinación del descriptor CC con otros descriptores como DESIRE y Spherical Spin Image, se logr´o un porcentaje promedio de reconocimiento de gestos del 90.27% y del 97.2 %, usando los mismos clasificadores mencionados previamente.