SUSAN清晰度评价函数在自动对焦中的应用

Application of SUSAN definition evaluation function in auto-focusing

为了满足自动对焦系统的实时性与抗噪性要求,提出了一种基于SUSAN算子的清晰度评价函数。该函数利用SUSAN边缘提取算子的算法简单、准确度高、抗噪性强的特点,对SUSAN算子边缘检测函数进行改进,将边缘点的USAN值的平方和作为清晰度评价值,并将其运用到自动对焦算法中。将该函数与几个经典的清晰度评价函数进行性能比较,实验结果表明:对于引入噪声前后的图像序列,基于SUSAN算子的清晰度评价函数均具有良好的单峰性、无偏性和较高的灵敏度;对于256×256的对焦窗口图片,该函数在TMS320C6416硬件平台上的运行时间仅为16 ms。该函数能够满足清晰度评价函数的单峰性、无偏性、高灵敏度等基本特性,同时具有良好的实时性与抗噪性。

In order to meet the real-time and anti-noise requirements of auto-focusing system, a definition e- valuation function based on SUSAN is proposed. This function is derived from the SUSAN edge detection algo- rithm, which makes the sum of squared edge point＇s USAN values as the definition evaluation value, and ap- plies to the auto-focusing system, making use of the high accuracy and strong anti-noise characteristics of SU- SAN algorithm. The definition evaluation function performance is compared between SUSAN function and the other classical functions. Experiment results show that the SUSAN definition evaluation function has the char- acteristics of clear single apex, good unbiased, high sensitivity for the image sequences before and after the in- troduction of noise. In addition, the running time of the function on the TMS320C6416 hardware platform is 15 ms. It can satisfy the definition evaluation function requirements of single apex, unbiased and sensitivity char- acteristics, as well as strong real time and anti-noise abilities.