一种面向水下无人航行器的实效图像增强算法

针对水下图像存在严重模糊、对比度低、噪声多等问题,根据小波的时频特性及多分辨率特点,提出一种基于小波变换的水下模糊图像增强算法。利用小波变换对RGB图像进行二层小波分解,把原图分解为高、低频子带,利用导向滤波算法估计低频子带上的照射分量并进行去除;利用软阈值算法对高频子带上的图像边轮廓信息进行去噪和增强处理;对水下图像进行小波逆构,并进行伽玛校正;最后利用改进的灰度世界算法对水下图像进行颜色校正。试验结果表明,使用文章中算法处理所得到图像的对比度及信噪比都较高,且清晰度较高,满足水下无人航行器的要求。

An Efficient Underwater Image Enhancement Algorithm for Unmanned Underwater Vehicle

Aiming at the current issues of underwater images, such as fuzzy, low contrast and noise, according to the time-frequency and multi-resolution characteristics of wavelet, an effective algorithm based on wavelet transform is proposed. The RGB images are decomposed to low frequency sub-band and high frequency sub-bands, the illumination component is estimated and removed using guided filter from the low frequency sub-band, and the high frequency sub-bands corresponding to the image edges are de-noised and enhanced by using soft-threshold algorithm. Then, the underwater image is reconstructed by wavelet transform, and the reconstructed image is adjusted by gamma algorithm to enhance the brightness of the image. Finally, the improved gray world algorithm is adopted to correct the color of the underwater images. The experimental results indicate that the proposed algorithm has many advantages, such as higher de-noising capability and excellent image contrast, which meets the requirements of the unmanned underwater vehicle.