Abstract: The aim of the work is to develop an algorithm for extracting local extrema of images with low computational complexity and high accuracy. The known algorithms of the search for local extrema based on non-maximum (or non-minimum) suppression have low computational complexity, but only strict maxima or strict minima are detected without errors. The morphological algorithms give accurate results, in which the extreme areas are formed by strict and non-strict extrema. However, it has a high computational complexity, separate process of the search for maxima and minima (iterative processing). In this paper, a new modified non-maximum suppression algorithm to find all local extrema on grayscale images is proposed. The essence of the algorithm is to search for single-pixel local extrema and regions of uniform brightness, comparing the values of their boundary pixels with the values of the corresponding pixels of adjacent regions by following: the region is a local maximum (minimum) if the values of all its boundary pixels are larger (or smaller) to the values of all adjacent pixels. The proposed algorithm allows to detect all single-pixel local extrema and extreme areas in images. Besides, the proposed algorithm in comparison with the morphological algorithm requires low computational complexity and reduces the processing time and the use of RAM.
Keywords: Local Extrema; Non-Maximum Suppression; Image Segmentation; Region Growing
| DOI: 10.17148/IJARCCE.2019.8901