小波变换在舰船图像尺寸测量方法中的应用

Application of wavelet transform in ship size measurements

分析小波阈值降噪和基于小波图像分割技术在舰船图像尺寸测量中的应用,在小波阈值降噪阶段,利用高斯小波对图像进行4层分解,逐层去除噪声的小幅值小波系数,达到去噪目的;在图像分割阶段,首先利用灰度直方图以及本文提出的像素邻域差值直方图构建二维直方图,然后对其进行最优一维投影;最后利用Haar小波对最优一维投影进行分解,求取图像最佳分割阈值。本文用提出的算法对舰船的主轴长度、主轴垂直方向长度、舰船面积以及周长4个尺度量进行测量。实验结果表明,小波变换在图像降噪和分割方面有诸多优势,而降噪和分割又是舰船尺寸测量的必须步骤,因此小波变换将在舰船图像尺寸测量中发挥越来越重要的作用。

This paper focus on the wavelet threshold de-noising and wavelet-based image segmentation technology in ship size measurement,in the wavelet threshold de-noising stage,Gaussian wavelet was used to decompose the image into four layers,and the noise with small wavelet coefficient is removed through all layers. In the image segmentation stage,gray histogram and the proposed pixel neighborhood difference histogram were used to construct a new two-dimensional histogram,in order to obtain the best image segmentation threshold,we use Haar wavelet to decompose the best projection of this two-dimensional histogram. Based on the algorithm proposed here,four features related to ship size were measured： length of ship spindle axis,length of axis vertical to spindle direction,ship area and perimeter.The experimental results show that wavelet transform has many advantages in image noise reduction and segmentation area,and the two procedure are base for ship size measurement,so wavelet transform will play an increasingly important role in ship measurement.