改进的先验去雾算法在海雾场景中的应用

Application of improved prior defoggingalgorithm in sea fog scene

为了提高智能船舶视觉系统对海雾图像的可视性,针对现有去雾算法对海雾场景处理时易造成天空区域严重失真,过饱和以及复原图偏暗等现象,提出一种改进先验的海雾去雾算法。由颜色衰减得到深度图,结合中值滤波补偿暗通道得到暗通道图,根据景深图和暗通道图设定阈值,分割出天空区域,分析天空区域的景深图获取大气光估计值;然后将暗通道与颜色衰减的介质传输图通过小波融合得到合成的介质传输图;为了优化去雾效果,对介质传输图设定上阈值与下阈值进行校正,再通过求解大气散射物理模型实现对图像清晰化处理;最后通过伽马校正优化色调偏暗的问题。通过对比实验验证,该方法应用在海雾场景下能够获得质量和视觉效果更好的去雾图像。

In order to improve the visibility of intelligent ship vision system to sea fog images,aiming at the phenomena such as severe distortion of sky area,oversaturation and darkening of the restoration image caused by existing defogging algorithms when dealing with sea fog scene,an improved priori algorithm for sea fog removal is proposed.The depth map is obtained through color attenuation,and the dark channel map is obtained by compensating the dark channel with the median filtering.According to the depth of field map and dark channel map,the threshold value is set to segment the sky area,and the estimated atmospheric light value is obtained by analyzing the depth of field map of the sky area.Then,dark channel and color-attenuated dielectric transmission image obtain synthetic medium transfer diagram by wavelet fusion.To optimize defogging effect,medium transfer diagram are set suprathreshold and L-threshold for correction.By solving atmospheric scattering physical model,realize sharp processing of image.Finally,the problem of dark tone is optimized by Gamma correction.Through comparison experiments,it is verified that this method can obtain better quality and visual effects of defogging images in sea fog scenes.