Summit Navigator: A Novel Approach for Local Maxima Extraction.

Publication Type:
Journal Article
Citation:
IEEE Trans Image Process, 2019
Issue Date:
2019-08-12
Filename Description Size
OCC-146903_AM.pdfAccepted Manuscript Version8.92 MB
Adobe PDF
Full metadata record
This paper presents a novel method, called the Summit Navigator, to effectively extract local maxima of an image histogram for multi-object segmentation of images. After smoothing with a moving average filter, the obtained histogram is analyzed, based on the data density and distribution to find the best observing location. An observability index for each initial peak is proposed to evaluate if it can be considered as dominant by using the calculated observing location. Recursive algorithms are then developed for peak searching and merging to remove any false detection of peaks that are located on one side of each mode. Experimental results demonstrated the advantages of the proposed approach in terms of accuracy and consistency in different reputable datasets.
Please use this identifier to cite or link to this item: