Edge Detection of Multispectral Images using Nonparametric Local Density Estimation

Ch. Theoharatos, V. Tsagaris, G. Economou, S. Fotopoulos, and V. Anastassopoulos (Greece)


Edge detection, multispectral, Landsat, SAR, density estimation, multivariate detection.


Detection of edges in multispectral images has been a challenging task in the research community over the past few years. In this work, a novel vector-based approach is adopted for edge detection in multichannel remotely sensed images. The discontinuity between homogeneous regions in the image is detected using the image density value estimated at the mean vector of the sliding window. The proposed algorithm is nonparametric, computationally simple to implement, providing us with dimensionality reduction in the multivariate feature space.

Important Links:

Go Back