【摘要】 针对暗通道先验除雾算法运算效率低以及复原图像中出现Halo效应等问题,提出一种基于形态学和高斯滤波的暗通道先验图像去雾算法。使用暗通道先验方法估计全局大气光值;对最小通道图采用形态学开操作进行滤波;采用高斯滤波进行平滑处理,降低开操作产生的像素间差异;运用膨胀运算消除相邻区域间的梯度,进而精确估计透射率;通过大气散射物理模型快速、准确地获取复原图像。实验结果表明,该算法不仅能够取得良好的除雾效果,得到细节丰富、对比度和亮度适宜的复原图像,而且大大降低了时间的复杂度,提高了运算效率。更多还原

【Abstract】 Aiming at the low efficiency of dark channel prior dehazing algorithm and the problem of halo in restored image, we propose a dark channel prior image dehazing algorithm based on morphology and Gaussian filtering. The dark channel prior method was used to estimate the global atmospheric light value, and then the minimum channel image was filtered by morphological opening operation. We used Gaussian filtering to smooth the image and reduce the difference between pixels. The gradient between adjacent regions was eliminated by dilation operation, and then the transmittance was estimated accurately. Through the physical model of atmospheric scattering, the restored image could be obtained quickly and accurately. The experimental results show that the algorithm can not only achieve good dehazing effect, get the restored image with rich details, appropriate contrast and brightness, but also greatly reduce the complexity of time and improve the efficiency of operation.更多还原