基于融合的夜间视频图像增强算法

夜晚条件下,能见度低,光线不足等问题导致夜间图像亮度通常很差而且亮度不均匀,从而影响图像质量。首先采集针对同一场景的白天图像,然后再将实时夜间视频进行高斯背景建模,提取出运动目标并将该夜间图像进行增强,对增强后的夜间图像和白天图像进行小波分解,然后用加权融合算法进行融合得到最终的背景图像,将最终的背景图像进行小波反变换,反变换以后的图像与提取出的目标叠加就得到增强后的图像。增强后的图像场景清晰,图像整体上自然、平滑、细腻,算法有效地提高了夜间图像的质量。

A Fusion based Night Image Enhancement Algorithm

Night images usually have low illumination and imbalanced illumination because of low visibility and insufficient light. The proposed algorithm firstly obtain day background images, then construct the real time night background using gaussion model which extract the moving target from the video and enhance the current night scene image using simultaneously. Both the night scene background image and day background image are decomposed using wavelet transform and then fused in wavelet domain with the proposed fusion algorithm to form the final background, moving target is just easily added into the final background and reverse wavelet transform is applied on it to obtain the final enhanced image. The enhanced image appears natural, smooth and neat and the proposed algorithm effectively improves image quality.