摘要
为了实现对行车视频中单幅雾霾图像的快速去雾,针对现有暗原色先验去雾方法在局部明亮区域存在透射率估算过低、算法效率低等问题,提出了一种改进的去雾方法。该方法在处理不同的像素点时,通过设定门限值,使得门限值范围内的区域保持原透射率不变,不在门限值范围内的区域使用边界条件限定,结合调整参数重新计算透射率,求出更准确的自适应透射率图。实验结果表明,这种方法修正了错误估算的透射率,克服了原算法在处理这些区域时产生的色彩失真,生成了高质量的无雾图像。
The existing dark colors prior defogging method is facing some problems that the estimates of transmittance were too low,low efficiency and so on,when being in the local bright area. In order to realize the rapid defogging for haze images in video traffic single,an improved method of dehazing is proposed. When the method in handling different pixel,by setting the threshold value,for the region in line with the threshold value within a range of maintaining the original transmission rate unchanged,boundary conditions are limited to areas that are not within the range of threshold values. Combined with the adjustment parameters to recalculate the transmittance,more accurate adaptive transmission diagram can be obtained. Result shows that after treatment with this method,effectively corrected the misjudgment of atmospheric light can be obtained. It eliminates the color distortion and the execution time of the algorithm is greatly reduced.
出处
《微型机与应用》
2017年第8期45-47,共3页
Microcomputer & Its Applications
关键词
图像去雾
暗通道先验
门限
透射率
大气光
image dehazing
dark channel prior
threshold
transmissivity
atmospheric light
