| 研究課題/領域番号 |
14F04041
|
|---|
| 研究機関 | 鹿児島大学 |
|---|
研究代表者 |
川崎 洋 鹿児島大学, 理工学研究科, 教授 (80361393)
|
|---|
| 研究分担者 |
VISENTINI SCARZANELLA Marco 鹿児島大学, 理工学研究科, 外国人特別研究員
|
|---|
| 研究期間 (年度) |
2014-04-25 – 2017-03-31
|
| キーワード | Computer Vision / 3D Reconstruction / Photometric stereo / Surgical navigation |
|---|
| 研究実績の概要 |
After the first 8 months of research, we have already touched three out of the four proposed tasks in the JSPS fellowship plan. First, we developed a system to calibrate position and characteristics of any light source. This method is practical and novel, and allows to understand the way how the light propagates according to the characteristics of the bulb. We have also shown that when taking into account the light propagation characteristics, it is possible to reconstruct the 3D shape of an object without moving the camera, but by simply moving the light source instead. The method for the first time in the literature it allows to recover the dense 3D shape of the object, which is the second achievement. Finally, we adapted our method for medical applications, and we have already performed our preliminary experiments on human tissue samples at Hiroshima University Hospital. Our overall system consists of a combination of traditional LED light together with a special laser light pattern integrated in a standard endoscope, which allows the surgeon to calculate the size of cancerous tissue. Other than the three achievements, we have found more common ground for collaboration on some of the existing projects in our laboratory, related to augmented reality. The scientific production has been fruitful, with 1 paper accepted and 3 more under review, and opportunities for patents. Overall, the plan is proceeding along the predicted lines and no change
of course is necessary.
|
| 現在までの達成度 (区分) |
現在までの達成度 (区分)
2: おおむね順調に進展している
理由
The fellowship plan contained the following four tasks: T1.Joint calibration for the camera-illumination-laser projector system, T2.Dense reconstruction from sparse 3D points, T3.BRDF estimation from scene structure, T4.Applications for intra-operative surgical guidance.
Regarng T1, we have completed the development of a system for joint calibration of light and camera characteristics without any special equipment. This will be soon submitted to a premium conference in computer vision and the task is completed. Significantly, we also realised how by modelling the light characteristics we are able to recover the real 3D dimensions of the object with photometric stereo, which was previously not possible. This result was used for T2, where we have developed a system for dense reconstruction using our method starting from a sparse structured light reconstruction. For T2 we also developed a dense reconstruction system based on structured light and image stereo suitable for multi-camera systems. The latter has been accepted for publication. The system will be finalised in the following months. While T3 will be tackled after T2's completion, we have already developed a prototype system for T4 that has been already tested on human tissue samples for measuring the size of tumours. 2 paper from T4 is under review, while another is ready for submission. Finally, aside from the plan we also developed novel solutions for augmented reality that are currently under review. In total, T1 is done, T2 and T4 are almost completed, 1 paper has been accepted and 4 are under review.
|
| 今後の研究の推進方策 |
The plan is proceeding according to the original fellowship plan and does not require any modifications at the moment. In terms of future plans, we will conduct further experiments on human tissue to test our prototype. This is also required in order to have sufficient data that can be used for a clinical publication, which would help increasing the impact of the project to a non-engineering audience. Similarly, while T1 is concluded, further experiments will be conducted in order to publish a more comprehensive study on our developed technique. Therefore, the theoretical aspects regarding T1,T2 and T4 have been largely completed and research activity until the end of the first fellowship year will be devoted to data collection, testing, and publication of more comprehensive studies. Work on T3 and eventual new applications on T4 will commence during the second year of fellowship.
|
