WSEAS Transactions on Signal Processing
Print ISSN: 1790-5052, E-ISSN: 2224-3488
Volume 15, 2019
Image Segmentation Based on Adaptive Mode Quantization and 2D Histograms Analysis
Authors: ,
Abstract: In the paper is presented a new method for segmentation of halftone images, based on adaptive quantization of the generalized one-dimensional histogram modes obtained through analysis of the oriented 2D histograms and double nonlinear grey level transform. Specific for the method is that it does not use thresholds defined through iterative analysis of the image histogram which are characteristic for the famous threshold segmentation methods. The new approach permits hierarchical image segmentation through reduction of the grey levels number in the initial hierarchical level. In each consecutive segmentation level the number of grey levels is increased, until the maximum is got. The method is illustrated by an example which shows higher segmentation accuracy, when compared to the well-known iterative Otsu method based on thresholds calculated through one-dimensional histogram analysis. The high accuracy and the low computational complexity of the presented method open new possibilities for real-time applications in the contemporary computer vision systems
Search Articles
Keywords: image segmentation, adaptive quantization of the one-dimensional histogram modes, oriented 2D histograms, multiple grey-level transform
Pages: 121-128
WSEAS Transactions on Signal Processing, ISSN / E-ISSN: 1790-5052 / 2224-3488, Volume 15, 2019, Art. #15