论文部分内容阅读
显微图像自动聚焦的关键在于设计一个高灵敏度聚焦函数。由于显微图像细节多寡不确定,传统的梯度函数对细节较少的图像的灵敏度不够高。针对该问题,提出了一种结合全局和局部灰度变化的VarGrad显微图像自动聚焦函数。根据显微图像的特点,VarGrad函数利用聚焦窗口将基于全局灰度变化的灰度方差函数与基于局部灰度变化的梯度函数有机结合,无论图像细节是否丰富,都呈现较高的灵敏度。实验利用两组细节丰富程度不同的外周血细胞图像序列对VarGrad函数进行了定量评估。实验结果表明,与几种典型的聚焦函数相比,在图像细节较丰富和图像细节较少两种情况下,VarGrad函数在清晰度比率、陡峭度和清晰度变化率3种灵敏度指标上均提高了30%以上。
The key to auto focus microscopic image lies in designing a high sensitivity focusing function. Due to the uncertainty of the details of the microscopic images, the traditional gradient function is not sensitive enough to the less detailed images. In response to this problem, a VarGrad auto-focus function combining global and local gray-scale changes was proposed. According to the characteristics of the microscopic image, the VarGrad function uses the focus window to combine the gray variance function based on the global gray level and the gradient function based on the local gray level. It shows high sensitivity whether the image detail is rich or not. The VarGrad function was quantitatively evaluated using two sets of peripheral blood cell images with different levels of detail. Experimental results show that compared with several typical focus functions, the VarGrad function improves on the three sensitivity indexes of sharpness ratio, steepness and sharpness change rate under the condition of rich image details and less image details More than 30%.