基于不变性特征的水下目标特征提取

针对水下成像环境的特殊性和复杂性，分析了在离散状态下比例因子对不变矩特征的影响，构造了基于区域矩的仿射变换不变量，以克服水下不确定因素给目标识别带来的困难，为了验证所提取特征的有效性，对球体、椭球体、三棱柱和四棱柱4类水下目标进行了特征提取试验．仿真试验结果表明，该方法在对简单背景水下图像的特征提取上能够取得较好的效果，可有效地克服水下图像灰度分布不均和环境不确定因素的干扰，实现了对水下目标的区分．

Underwater target feature extraction based on invariant features

This paper analyses the influence of scale factors on moment invariant features in discrete conditions,constructing affine transformation invariants based on regional moments in order to eliminate effects caused by the particularity of underwater channels and the complexity of the imaging environment.A new method for corner extraction is also proposed.In simulations,four different objects were selected and the features were extracted with the new method.The experimental results demonstrate that this approach can effectively reduce interference caused by non-uniform gray levels and uncertain environmental factors.The proposed method is valid for feature extraction in underwater images with simple backgrounds,showing distinctions between different objects.