中巴地球资源卫星图像中伪“同谱异物”区分方法研究

Research on the Method of Differentiating False"Different Objects Same Image"in the CBERS lmage

通过对中巴资源卫星图像中存在的伪"同谱异物"现象的分析,提出区分这两类地物的图像处理方法。首先对图像进行傅立叶变换,在频率域图像中针对噪声特点,分别采用楔形、矩形和圆形掩模滤波方法,消除图像条带噪声和周期性噪声,再用中值滤波去除随机噪声;其次对各波段图像进行波段配准和几何精校正;最后进行直方图匹配和拉伸处理,增强图像清晰度和分辨力。得到的图像中"同谱异物"现象消失,实际分类结果证实该处理方法的有效性和实用性。

By analyzing the false'different objects same image'phenomenon in the CBERS image,this paper proposes a method to distinguish the two objects.Firstly,the Fourier transform is done to the image.And the Wedge Mask,Rectangular Mask and Circular Mask filtering methods are used respectively to remove the stripe noise and periodicity noise according to the characteristic of the noise of the im- age in the frequency domain.Then the random noise is removed by the median filter.Secondly,the band match and geometric accurate correction are done to the image of every band.At last,the histogram match and LUT stretch are carried out to enhance the definition and resolving power.The'different objects same image'phenomenon disappears in the final image.The actual sorting result proves the validity and practicability of this method.