The binarization process is the preliminary and most significant phase of the document image analysis applications. A hybrid approach based on the merger of Salp swarm algorithm and the chaos theory is introduced. The proposed hybrid approach has been used to evaluate their ability and precision in the clustering process. It is revealed how Salp can operate to find automatically the centroid of a defined number of clusters using K-means objective function. Several different chaotic maps are integrated to adjust the behavior of the Salps by calibrating their random numbers. The efficiency of the proposed chaotic Salp swarm algorithm is empirically verified on the Document Image Binarization Contest H-DIBCO 2016 dataset. A comparison made between the proposed approach and some of the state-of-the-art methods in terms of F-Measure, Peak Signal to Noise Ratio and pseudo-F-Measure. In addition, Geometric-mean pixel accuracy, Distance Reciprocal Distortion Metric, Negative Rate Metric and Misclassification Penalty Metric are shown and discussed.
K-means, Salp swarm Algorithm, Optimization, Clustering, H-DIBCO 2016.
Mohamed ABD ELFATTAH, Mohamed ABD ELFATTAH, Sherihan ABUELENIN, "A Hybrid Swarm Optimization Approach for Document Binarization", Studies in Informatics and Control, ISSN 1220-1766, vol. 28(1), pp. 65-76, 2019. https://doi.org/10.24846/v28i1y201907
All articles published by Studies in Informatics and Control (SIC) are published in accordance with the terms and conditions of the Creative Commons Attribution License (CC BY-NC-SA 4.0).