夜间视频图像增强是夜视成像的关键技术, 而由于夜晚环境照度低, 存在路灯等局部亮点, 限于器材原因, 易造成夜视图像“局部饱和, 但灯下黑”的特点。根据城镇环境下极低照度CMOS视频成像的特点, 基于拉普拉斯金字塔重构算法, 将不同曝光参数下的同一场景图像进行分解, 通过熵和平均梯度数值筛选图像细节, 重构出细节丰富的场景。实验表明, 该方法在处理夜视图像时可以有效还原图像细节, 获得高动态范围图像。

Night video image enhancement is a key technology of night vision imaging. Ambient illumination is low at night, and there are street lamps and other local highlights. For these equipments, the night vision images are easy to have the features of local saturation but dark under the lamp. According to CMOS video imaging characteristics with low illuminance in urban situation, based on Laplacian pyramid reconstruction algorithm, the same scene images with different exposure parameters are decomposed, and the scene with details is reconstructed through choosing image details on the basis of entropy and average gradient values. Experiments show that the method can be used to restore the image details of night vision images effectively and high dynamic range images are obtained.