Computer Vision
[1] Development of Multi-channel Camera Systems
・Development of a multi-channel camera
A multi-channel camera with more than three spectral bands provides an effective tool for estimating the complicated shapes of the surface and illuminant spectral functions in natural scenes.
・Material Classification for Printed Circuit Boards
Multi-channel images are useful for classifying object materials on a raw circuit board based on surface-spectral reflectance. The performance is better than the RGB-based method using a normal color camera.
[2] Material Classification by Polarization
The problem of distinguishing material of object surfaces from image data is an essential problem in many fields including object recognition. This study aims to distinguish dielectric and metal material surfaces by the map of the degree of polarization.
[3] Estimation of Reflection Properties from Image Data
The reflection properties of an object surface depend on the object's material composition. We are interested in determining the reflection model from the image data of a real object.