A Fast 2-D Otsu lung tissue image segmentation algorithm based on improved PSO
Yanqiao Zhao,Xiaoyang Yu,Haibin Wu,Yong Zhou,Xiaoming Sun,Shuang Yu,Shuchun Yu,He Liu
DOI: https://doi.org/10.1016/j.micpro.2020.103527
IF: 3.503
2021-02-01
Microprocessors and Microsystems
Abstract:<p>In order to reduce the time of lung tissue image segmentation, we proposed a fast 2-D Otsu lung tissue image segmentation algorithm based on improved PSO. Firstly, in the 2-D Otsu algorithm, the search scope of 2-D gray threshold is limited in a long and narrow region, which is the neighbourhood of the diagonal from the third region to the first region of the 2-D gray histogram, and the search scope and the computation are reduced, operation speed is improved. Secondly, In the PSO algorithm, the position of the particles is adjusted during iterating based on the principle of symmetric disposition, so as to avoid PSO falling into local optimal solution and improves the accuracy of threshold searching. Finally, a lung CT image with 1280×960 resolutions is segmented by our algorithm and other traditional algorithms, and a comparison is given. The segmentation threshold of our method is 85, the difference is less than 5 comparing with that of other traditional algorithms, and it shows that our method has almost the same searching precision as the traditional algorithms. The time cost is only 162ms, which is far less than the traditional algorithms, and it shows that our method improve the segmentation speed. It can be concluded that our method can not only satisfy the requirement of segmentation precision, but also meet the requirement of operation speed.</p>
computer science, theory & methods,engineering, electrical & electronic, hardware & architecture