Picture of person typing on laptop with programming code visible on the laptop screen

World class computing and information science research at Strathclyde...

The Strathprints institutional repository is a digital archive of University of Strathclyde's Open Access research outputs. Strathprints provides access to thousands of Open Access research papers by University of Strathclyde researchers, including by researchers from the Department of Computer & Information Sciences involved in mathematically structured programming, similarity and metric search, computer security, software systems, combinatronics and digital health.

The Department also includes the iSchool Research Group, which performs leading research into socio-technical phenomena and topics such as information retrieval and information seeking behaviour.

Explore

Pixel clustering and hyperspectral image segmentation for ocean colour remote sensing

Zeng, Xuexing and Ren, Jinchang and McKee, David and Lavender, Samantha and Marshall, Stephen (2012) Pixel clustering and hyperspectral image segmentation for ocean colour remote sensing. In: Hyperspectral Imaging Conference, 2012-05-15 - 2012-05-16.

Full text not available in this repository. Request a copy from the Strathclyde author

Abstract

Hyperspectral dataset classification is a basic task for ocean colour remote sensing [McKee et al., 2007], [Borengasser et al., 2004]. In this paper, region growing is proposed to classify hyperspectral dataset [Adams et al., 1994]. Considering that it is difficult to select seeds, we select 20 by 20 uniformly distributed seeds for region growing. The region will grow from the seed by adding its 4-connected neighbours that is most similar with mean value vector. The Euclidean distance is used to measure the similarity between pixels. If the Euclidean distance between mean value vector and neighbour of seed is smaller than the threshold, this neighbour is considered that it is similar with this grown region, and this neighbour will be added to this growing region. Good classification results can be obtained by simply adjusting similarity threshold to specify the pixel similarity to preserve more or less details in the segmented results. With another parameter: size threshold for post- processing, the results can be further refined. One band sample of dataset and the classification results with threshold: 0.001 are shown in Figure 1, and more results will be presented in the full paper.