Image clarityis an important index to measure image quality. Clarity evaluate accurately has significant meanings for image processing system. In order to improve the sensitivity and noise immunity , this paper proposed a no reference image clarity evaluation method based on the point acutance method and squared gradient method. Firstly , image segmentation was fulfilled by introduced threshold and removed false edges. Thenthe image was divided intoflat zone and edge zone. Secondly , point acutance method and square gradient method were employed on the flat and edge zone respectively. Lastly , the image clarity is obtained through the weighted sum of flat sharpness and edge sharpness. Simulation experiment's results verified that the improved method reflects the advantages of point acutance and squared gradient. It has clear single apex, good unbiased, better sensitivity and noise immunity,which is able to evaluate the clarity of image accurately and stably. The improved method is suitable for evaluating blur image clarity, auto-focusing in actual optical image systems and comparing clarity of different size images.%图像清晰度是衡量图像质量优劣的重要指标,准确、高效地进行图像清晰度评价对于图像处理系统具有重要意义.为提高图像清晰度评价方法的灵敏度和抗噪性,提出了一种结合点锐度和平方梯度的无参考图像清晰度评价方法.该方法首先通过引入高低阈值和去伪边处理实现图像分割,得到图像平坦区和边缘区;其次对图像平坦区和边缘区分别应用点锐度法和归一化的平方梯度法计算清晰度;最后将二者加权求和得到图像清晰度.仿真实验结果表明,该方法体现了点锐度法和平方梯度法的优势,单峰性强、无偏性好、灵敏度高、抗噪性好,能够准确稳定地评价图像清晰度.该方法可用于模糊图像的清晰度评价、实际光学系统的自动调焦和不同尺寸图像的清晰度比较.
展开▼
