1P1-E04 Transparent Object Detection Using Color Image and Laser Reflectance Image for Mobile Manipulator(3D Measurement/Sensor Fusion)

概要

論文の詳細を見る
It is a significant challenge to detect or model a transparent object using conventional visual sensors. Here, we show that the characteristics of transparent objects under different lighting conditions (normal light and laser light) can be exploited for improved detection accuracy. Specifically, we analyze the object characteristics and combine them into detection parameters. We propose a method to specify candidate areas on the basis of highlight spots in the color image and then use these areas as bounding rectangle inputs for GrabCut. Thereafter, segmentation by GrabCut is applied to depth images and laser reflectance intensity images. Next, the transparency of the object is determined qualitatively by comparison of the segmentation results for the depth image with those for the reflectance intensity image. Bottle-shaped transparent objects are correctly detected within 1 m of the visual sensor using the proposed method. Evaluation was carried out for different shapes, materials, distances, and backgrounds.
一般社団法人日本機械学会の論文
2011-05-26