【摘要】 提出一种新的图像清晰度评价函数,该函数对经典Sobel边缘检测算子进行了改进,采用八方向模板来检测梯度值,在一定程度上能有效地解决传统评价算子梯度计算中方位局限的缺点,同时将标准差作为阈值引入到函数中,可以消除噪声的影响。通过两组仿真实验对该算法和其他评价算法的聚焦性能进行了比较,结果表明该算法可行,具有单峰性强、无偏性好、灵敏度高和抗噪性强等特点。更多还原

【Abstract】 In this paper,a clarity-evaluation function was researched.An improved eight models Sobel function was proposed,to some extent,the problem of the limited orientations in other clarity-evaluation functions was solved effectively.Standard deviation as a threshold parameter was introduced to reduce the influence of noise.In emulation experiment,this function was compared with several other evaluation algorithms.The two emulation results show that this function has the characteristics of powerful single apex,high sensitive and strong anti-nosie ability.更多还原