亳州学院教学质量工程项目精品开放课程:模拟电子技术(2019gxkc05)。
针对单尺度 Retinex(SSR)图像去雾的算法效果不佳,探讨在图像去雾处理中多尺度 Retinex (MSR)算法去除浓雾、厚 雾的问题,并在 MSR 基础上进一步对图像进行带颜色复原的 MSR(MSRCR)的处理。 首先采用二维傅立叶变换得到高斯滤波 函数,同时由逆傅立叶变换、卷积积分计算出图像的入射分量,再通过算法计算出图像的反射分量,最后通过低通滤波器得到 图像的高频分量后还原图像。 实验结果显示:与 SSR 和 MSR 算法相比,MSRCR 算法能更好地去除图片中的浓雾、厚雾的影 响,提高带雾图片处理后的清晰度且效果明显。
Due to the poor effect of single-scale Retinex (SSR) image defogging algorithmꎬ this paper mainly discusses an algorithm based on multi-scale Retinex image defogging. Aiming at the defect that the dense and thick fog cannot be re ̄ moved in the defogging algorithm based on the multi-scale Retinex (MSR) imageꎬ we further perform MSRCR (MSR with color restoration) on the image based on the MSR. We first use a two-dimensional Fourier transform to obtain a Gaussian filter function. At the same timeꎬ the incident component of the image is calculated by inverse Fourier transform and con ̄ volution integralꎬ and then the reflection component of the image is calculated by the algorithmꎬ and finally the high fre ̄ quency component of the image is obtained by low-pass filtering. Compared with the SSR and MSR algorithmsꎬ experimen ̄ tal results show that the method can better remove the effects of dense and thick fog in the pictureꎬ improve the clarity of the treated picture effectively.
张晓东1 ,秦娟娟1∗, 贾仲仲2.多尺度 Retinex 在图像去雾算法中的应用研究[J].西昌学院学报(自然科学版),2021,35(3):60-65.