If you are an undergraduate or a masters student interested to do your final project in any of the below topics or you would like to do an internship at TNO, please feel free to contact me.
Topics of interest:
- Segmentation and superpixels
- Image classification of historical buildings
- Comparison of image classification using COSFIRE filters versus conventional key point detectors (e.g. SIFT, SURF, etc …)
- Development of a smart phone app for image classification
- Computer vision: including object recognition, contour detection, image classification, and segmentation
- Parallel implementation on a GPU of the COSFIRE (pattern recognition) approach. More information is given here.
- Machine vision: visual quality inspection
- Predictive analysis of non-visual type of data
- Pattern recognition in time series data
Moreover, the following is a project available in collaboration with TNO, in the field of Photogrammetry
Photogrammetry is a computational technique to reconstruct a 3D structure from a collection of pictures (photos, video). Typical applications of photogrammetry can be found in archeology and land/site measurement. In archeology, photogrammetry is a popular tool for creating a 3D structure of a landscape by taking aerial pictures with such a a high resolution that is possible to uncover very fine details like foundations, roads that are unseen from a ground level or a single aerial picture. With the arrival of relatively cheap methods for taking aerial pictures: RPAS (remote piloted aerial systems), UAS (unmanned aerial systems), KAP (kite aerial photography), BAP (balloon aerial photography), PAP (pole aerial photography) the popularity of photogrammetry has grown in the last decade.
A part of the province Groningen suffers from earth quakes related to the drilling and extraction of gas from the earth in this area. While these earthquakes result in house damage: cracks and displacements, the relation between an earthquake and the impact on a house is still not clear. In fact this is a research topic in a project being carried out for the NAM at this moment.
The goal of this project is to find out if there is any added value in tools such as photogrammetry for detection of changes in houses, landscapes as a result of an earth quake. It is the goal to find out if this is possible from a technical perspective, but also investigate the possibilities for practical application in the field.
- What is the state of the art in Photogrammetry?
- Are there any known applications of Photogrammetry applied to buildings?
- What is the required resolution that is expected to be usefull for detection changes in building structure?
- Is it possible to achieve this resolution using photogrammetry? What are requirements for camera’s etc.
- Is it possible to design a workflow/procedure for taking pictures after a workflow?
- How does the result compare to current methods (manual inspection, and a single image)
- How can the positions and viewing angles of the camera, which are automatically determined by 3D Photogrammetry, be used to achieve better panoramic stitching techniques?