传统梯度类自动调焦评价函数对不同清晰度的序列图像不能兼顾灵敏度高和可调焦范围广两种特性, 因此提出了基于人类视觉系统(HVS)算子的自适应清晰度自动调焦评价函数。检测图像各方向上的边缘, 计算轻微离焦图像函数值; 当图像深度离焦时, 划分子区域降低采样率后进行清晰度评价。仿真实验表明本文提出的评价函数, 轻微离焦位置灵敏度高, 对深度离焦区域图像加均值为 0, 方差为 0.001,0.005,0.01的高斯噪声后函数曲线几乎不变, 深度离焦区域内可调焦范围大。该评价函数具有灵敏度高、可调焦范围大、抗噪性能良好等特性, 更适合粗精结合的搜索算法使用。

The traditional gradient auto-focusing evaluation function can not take into account the high sensitivity and the wide focus range for different clarity sequences. Therefore, a self-adaptive clarity auto-focusing evaluation function based on human visual system (HVS) operator is proposed. Detecting the edge of each direction of the image, calculating the value of the minor defocused image; when the image is depth defocused, the sub-region reduces the sampling rate and then performs the sharpness evaluation. The simulation results show that, the function sensitivity in the mild defocusing position is high, the function curve is almost invariant in the depth defocus region after the Gaussian noise which mean value is 0, the variance is 0.001, 0.005, 0.01 is added to the image, the wide depth defocus region can be focused. The evaluation function has the characteristics of high sensitivity, large focus range, good anti-noise performance and more suitable for the rough-precise searching algorithm.