半导体激光夜视图像中的可逆数据挖掘

Reversible data mining method of semiconductor laser night vision image

半导体激光夜视图像采用外用激光照明器,在全黑通条件下进行采集,存在大量干扰数据,一些有后期应用价值的可逆数据在夜视下存在较大缺陷,导致挖掘特征无法定位,为此提出了逆推聚类挖掘理论的半导体激光夜视图像中可逆数据挖掘方法,将半导体激光夜视图像指标特征值进行合理的处理再对其进行分类。依据激光夜视图像可逆数据特有的样本模糊像素可衡量特征,引入模糊像素差异迭代聚类挖掘方法,实现半导体激光夜视图像中可逆数据挖掘。测试实验结果表明,这种方法对半导体激光夜视图像中可逆数据挖掘耗时和准确性都有较为明显的改善,具有较高的应用价值。

In the black condition, semiconductor laser night vision image using external illuminator to collect, there are a large mount of interference data. And some reversible data which have late application value have big de- fects, makes the mining feature cannot be located. Therefore, a reversible data mining method of semiconductor laser night vision image based on reverse clustering mining theory was proposed, which can process rationally and classify the index feature of semiconductor laser night vision image. Bring in the fuzzy pixel difference iterative clustering min- ing method according to the measurable feature of its special sample fuzzy pixel to realize reversible data mining in sem- iconductor laser night vision image. The test results show that this method improved time saving and accuracy, and it has high application value.