作者: I-Ting Fang , Bolan Jiang , Aditya Sharma , Disha Ahuja
DOI:
关键词: Pattern recognition 、 Cluster analysis 、 Mathematics 、 Artificial intelligence 、 Image segmentation 、 Pixel 、 Edge orientation 、 Image separation 、 Image based 、 Entropy (information theory) 、 Entropy estimation
摘要: Entropy based image segmentation determines entropy values for pixels in an on intensity or edge orientation. One more threshold are determined as a fraction of the distribution over image. For example, high and/or low thresholds may be generated to identify regions associated with trees sky, respectively. The compared threshold(s) from which within can segmented. Intensity has no structural information, and thus, proximity clustering pruning points is performed. A mask applied segmented remove image, useful in, e.g., objection recognition processes. Additionally, separate buildings identified using orientation pruning.