基于改进麻雀搜索算法的森林火灾图像多阈值分割

为了减少森林火灾对人们的生活和生态环境带来巨大的损失和破坏,采用图像处理技术对森林火灾进行精准定位和预判,可以有效地降低火灾的扩大和蔓延。针对火灾图像大津法分割算法计算量大、运行时间长和阈值选取不够准确导致分割精度不高等缺点,提出精英反向学习-莱维飞行策略的麻雀搜索算法,并将其有效地应用到指数熵多阈值图像的分割中进行寻优,通过研究最佳阈值对森林火灾图像进行合理的分割,并与其他三种指数熵多阈值图像分割算法进行了对比分析。结果表明:改进麻雀搜索算法的森林火灾图像多阈值分割技术能够及时获得火灾图像分割的最佳阈值,其分割的准确性、实时性和抗噪性均明显优于现有的灰狼算法、粒子群算法和鲸鱼算法,能够为图像处理的工程应用提供一种较好的阈值分割技术。

Multi-threshold Segmentation of Forest Fire Image Based on Improved Sparrow Search Algorithm

In order to reduce the huge loss and destruction caused by forest fires to people''s lives and the ecological environment, the use of image processing technology to accurately locate and predict forest fires can effectively reduce the expansion and spread of fires. Aiming at the shortcomings of the fire image Otsu''s method segmentation method, such as large amount of calculation, long running time and insufficient threshold selection, which results in low segmentation accuracy, the sparrow search algorithm of elite opposition-based learning- Lévy flight strategy was proposed, and it was effectively applied to the exponential entropy multi-threshold image In the segmentation, the optimization was performed, and the forest fire image was reasonably segmented by studying the optimal threshold, and compared with the other three exponential entropy multi-threshold image segmentation algorithms. The results show that the multi-threshold segmentation technology of forest fire images with improved sparrow search algorithm can obtain the best threshold for fire image segmentation in time, and its segmentation accuracy, real-time performance and noise resistance are significantly better than the existing gray wolf algorithm, particle swarm algorithm and the whale algorithm can provide a better threshold segmentation technology for image processing engineering applications.