森林防火图像识别中的光照干扰消除仿真分析

Light Interference Elimination Simulation of Forest Fire Image Recognition

在物联网视觉识别技术在森林火灾报警的研究中,由于光照强度的随机性无法控制,光强度变化对图像特征采集带来了较大的干扰,对火焰像素与周围像素的差异化关联程度下降。传统的模式识别方法在光照干扰下,会由于火焰特征关联度的下降,造成火灾误识别。提出了一种空间域滤波的物联网视觉中森林火灾识别光照消除方法。利用空间域滤波方法,消除由光照造成的火焰图像噪声,提高了图像的分辨率,为森林防火提供了准确的数据源。针对特征关联下降的问题,引入线性最优分类平面,对干扰下的火焰特征进行分类,从而实现森林火灾识别。实验结果表明,上述算法可以有效消除物联网视觉中光照干扰,有效提高了森林火灾识别准确率。

Iot vision recognition technology application in forest fire alarm,the light intensity of randomness is unable to control,light intensity changes brought greater interference characteristics of image acquisition,on the degree of differentiation associated flame pixels and the pixels around the drop.Under the light interference,the traditional methods of pattern recognition will be reduced due to the characteristics of correlation degree,causing fire recognition by mistake.Presents a spatial domain filtering of the Intemet of things vision light elimination method for identifying the forest fire.Using the spatial domain filtering method to eliminate noise caused by light flame images,improve image resolution,for forest fire prevention provides accurate data source.In view of the characteristics of the decline of the association,into linear optimal classification plane,classifying under the interference of flame characteristics,so as to realize the forest fire.Experimental results show that the algorithm can effectively eliminate the Internet of things vision of light interference,improve forest fire recognition accuracy.