一种基于监控系统的图像清晰度检测算法

视频图像的清晰度是衡量视频监控系统的重要指标。通过比较灰度变化函数、频谱函数、方差梯度函数、灰度熵函数评价图像清晰度的优劣,提出了一种新的清晰度评价函数。通过测试大量的图片和监控视频,对改进后的算法和其他检测算法的性能进行了比较,结果表明,提出的检测算法具有单峰性强、灵敏度高、无偏好性、信噪比高等特点,能够自动实时有效地完成对视频图像清晰度的检测。

A Method to Detect the Definition of Video Image based on Video Surveillance

The definition of the video image is an important indicator to measure video surveillance system. In the paper it is compared to use Gray Level Variance, Frequency, Variance Gradient, Gray Entropy as evaluation functions for their advantages and disadvantages, it is proposed to a new method evaluate video Image definition. This function was compared with severl other evalution algorithms by testing a large number of pictures and surveillance videos. The results shows that this function has characteristics of powerful single apex, high sensitive, no preference and high Signal-to-Noise as well as this function can effectively to detect the definition of video image in real time.