基于机器学习参数选择的交通图像编码方法

将图像压缩与ITS交通路况图像电传视讯相结合,提出了一种基于机器学习参数选择的多项式拟合图像压缩编码方法.我们定义了两类指标来度量数据扫描方法对多项式拟合方法的影响;还研究了用机器学习方法选择得到的参数对图像扫描数据进行单调化处理;进而研究了用多项式拟合预处理数据的方法进行图像数据压缩.该法简单方便、快速高效,并已针对小幅复杂交通路况图像在中低信噪比取得了良好的结果.该法还可以推广到感兴趣区位于图像中心的一类图像.

A Traffic Image Code Method based on Machine Learning Parameter Choice

Combined image compression with traffic image transmission in ITS,this paper presents a new image compression method of polynomial approach which based on machine learning parameter choice.We defined two species of index to measure the stability of the image scanning ways.We also researched how to get the parameter which is produced by machine learning,and how to use the parameter to monotonize the image scanning data.Moreover,we compressed the scanned data using the polynominal approach method.The most merit of our method is facility and high efficiency.We have gotten good results aimed at the small complex traffic image,especially in middle and low signal-to-noise ratio.Our method also can be popularize to a kind of image which interest area is locating in the center of the image.