基于变粒度反馈认知的全景图像太阳方位研究

On solar azimuth in panoramic image based on variable granularity feedback cognition

文章采用改进的多尺度小波变换方法,获取太阳光斑不同尺度下的各类特征信息,并结合人类认知机理,构建了多类训练样本的不确定信息认知对象的仿真反馈认知智能机制与计算模型。通过太阳光斑的认知智能决策系统以及太阳光斑认知误差的定义,给出了一种待认知光斑不确定认知结果评价规则和相似光斑的变粒度反馈对比决策系统,准确获取了太阳光斑方位坐标以达到实际应用中快速性和准确性的平衡。

The improved multi-scale wavelet transform method is applied to getting all kinds of characteristic information of sun spots in different sizes. Combined with human cognition mechanism, a simulated feedback cognitive intelligent mechanism and computing model of uncertain information cognition objects of multiple training samples is constructed. Through the sun spot cognitive intelligent decision-making system and the definition of the sun spot cognitive errors, the evaluation rules of uncertain cognitive outcome of the sun spot and the variable granularity feedback and decision-making system for similar sun spots are given. The azimuth coordinates of the sun spot are obtained accurately, thus achieving the balance between the speed and accuracy of the application.