Determination of the impact of RGB points cloud attribute quality on color-based segmentation process
Journal Title: Bulletin of the Military University of Technology - Year 2015, Vol 64, Issue 2
Abstract
The article presents the results of research on the effect that radiometric quality of point cloud RGB attributes have on color-based segmentation. In the research, a point cloud with a resolution of 5 mm, received from FAROARO Photon 120 scanner, described the fragment of an office’s room and color images were taken by various digital cameras. The images were acquired by SLR Nikon D3X, and SLR Canon D200 integrated with the laser scanner, compact camera Panasonic TZ-30 and a mobile phone digital camera. Color information from images was spatially related to point cloud in FAROARO Scene software. The color-based segmentation of testing data was performed with the use of a developed application named “RGB Segmentation”. The application was based on public Point Cloud Libraries (PCL) and allowed to extract subsets of points fulfilling the criteria of segmentation from the source point cloud using region growing method.Using the developed application, the segmentation of four tested point clouds containing different RGB attributes from various images was performed. Evaluation of segmentation process was performed based on comparison of segments acquired using the developed application and extracted manually by an operator. The following items were compared: the number of obtained segments, the number of correctly identified objects and the correctness of segmentation process. The best correctness of segmentation and most identified objects were obtained using the data with RGB attribute from Nikon D3X images. Based on the results it was found that quality of RGB attributes of point cloud had impact only on the number of identified objects. In case of correctness of the segmentation, as well as its error no apparent relationship between the quality of color information and the result of the process was found.[b]Keywords[/b]: terrestrial laser scanning, color-based segmentation, RGB attribute, region growing method, digital images, points cloud
Authors and Affiliations
Bartłomiej Kraszewski
Keywords
Related Articles
- EP ID EP62772
- DOI -
- Views 92
- Downloads 0
Analysis of crack and strain states of high performance reinforced fiber concrete beams
The article presents the crack and strain states analysis of the reinforced high performance concrete beams with steel and polypropylene fibre. The beams were bent in the 4-point model. Research was carried out for three...
Testing the Homogeneity of Sequences of Multivariate Random Variables
The paper discusses the method of examining the homogeneity of sequences of random variables including testing the normality of multivariate random variables as well as testing the equality of many expected values and te...
Tłumienie dźwięku hełmofonu czołgisty
Załoga pojazdu bojowego, która jest narażona na hałas związany z pracą silnika, jak i z wystrzałami z broni palnej, ma na swoim wyposażeniu hełmofony. W artykule przedstawiono wyniki badań ograniczania hałasu przez hełmo...
Planowanie osłony przeciwlotniczej obiektu obszarowego za pomocą środków bardzo krótkiego zasięgu
W pracy zaprezentowano heurystyczną metodę planowania rozmieszczenia przeciwlotniczych zestawów rakietowo-artyleryjskich bardzo krótkiego zasięgu (VSHORAD) wokół osłanianego obiektu obszarowego. Jako miarę oceny jakości...
Zastosowanie logiki czterowartościowej w procesie wnioskowania w systemach diagnostycznych
W artykule zaprezentowano zastosowanie logiki czterowartościowej w procesie wnioskowania w systemach diagnostycznych. W tym celu przedstawiono i zdefiniowano strukturę inteligentnego systemu diagnostycznego DIAG 2. Opisa...