2005 Fiscal Year Final Research Report Summary
A Study on Fast and Robust Visual Sensing and Control Methods
| Project/Area Number |
16300049
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (B)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
Perception information processing/Intelligent robotics
|
|---|
| Research Institution | HOKKAIDO UNIVERSITY |
|---|
Principal Investigator |
KANEKO Shunichi Hokkaido University, Institute of Information Science and Technology, Professor, 大学院・情報科学研究科, 教授 (50134789)
|
|---|
| Project Period (FY) |
2004 – 2005
|
| Keywords | robustness / image registration / computation cost / occlusion / depth |
|---|
| Research Abstract |
In this study, matching or registration algorithms for the problems of ill-conditions, such as high-lighting, illumination fluctuation, complex background, occlusion, and defects, which often occurred in the real-world environment, could be developed. Two- and three-dimensional scenes were managed in the experiments. Especially, the robustness of the algorithms were considered in their design. In detail, Orientation code entropy based on Orientation code matching (OCM) was developed for practical robotic jobs, and then it was applied to realize a learning-based robot vision system. Next, color information and depth information were integrated into one representation and a fast and robust searching algorithm was realized for the color and depth scenes. It was applied to a tracking system of real objects. Last, for the problems including occlusion and complex background, Depth aspect method was developed for solving them by using local aspects of their depth information. It could applied to a real robot system for doing assembly tasks in the real world.
|
