基于聚焦曲面的快速聚焦算法的研究
|
摘要
在大倍率监控摄像机中,聚焦速度对摄像机有着重要的影响。为了提高聚焦速度和聚焦精度,通过对聚焦曲线的分析从而获得聚焦曲面,利用聚焦曲面,在变倍或者移动的同时进行跟焦处理,结合数字图像处理的方法,可以快速实现聚焦。为减小计算量,对分辨率为3840×2160的输出数据进行抽稀放缩,结合放缩后图像特点,使用快速DCT变换进行清晰度评价,从而实现局部小范围精确聚集。基于4K分辨率的33倍光学变焦网络摄像机对提出的算法进行验证,实验结果表明在保证画面清晰的跟焦处理结束后3.7秒内即可完成聚焦。
Focusing speed plays an important role in high power surveillance cameras.In order to improve the focusing speed and focusing accuracy,the focusing surface is obtained by analyzing the focusing curve.Focusing can be done while zooming or moving by using this focusing surface.With the method of digital image processing,focusing can be realized quickly.To reduce the computational complexity,the output image with resolution of 3840×2160 is scaled and diluted.With the characteristics of scaled image,the fast DCT transformation is used to evaluate the scaled image,so that local small-scale accurate focusing can be achieved.The proposed algorithm is validated by a 33 x optical zoom network camera with 4 K resolution.The experimental results show that the focusing can be completed within 3.7 seconds after focus tracking under the condition that the picture is clear.
Focusing speed plays an important role in high power surveillance cameras.In order to improve the focusing speed and focusing accuracy,the focusing surface is obtained by analyzing the focusing curve.Focusing can be done while zooming or moving by using this focusing surface.With the method of digital image processing,focusing can be realized quickly.To reduce the computational complexity,the output image with resolution of 3840×2160 is scaled and diluted.With the characteristics of scaled image,the fast DCT transformation is used to evaluate the scaled image,so that local small-scale accurate focusing can be achieved.The proposed algorithm is validated by a 33 x optical zoom network camera with 4 K resolution.The experimental results show that the focusing can be completed within 3.7 seconds after focus tracking under the condition that the picture is clear.
引文
[1]林忠,黄陈蓉,卢阿丽.应用于大变倍监控摄像机的电动变焦跟踪[J].应用光学,2015,36(5):684-690.
[2]蒋涛,左昉,王灵国.严重离焦状态下的自动聚焦实现[J].激光技术,2015,39(6):794-797.
[3]Lyu D,Xia H,Wang C. Research on the Effect of Image Size on Real-Time Performance of Robot Vision Positioning[J]. EURASIP Journal on Image and Video Processing,2018(1):112.
[4]罗钧,孙力,闵志盛.变焦跟踪曲线在对焦中的应用[J].光学精密工程,2011,19(10):2326-2332.
[5]Liu S,Liu M,Yang Z. An Image Auto-Focusing Algorithm for Industrial Image Measurement[J]. EURASIP Journal on Advances in Signal Processing,2016(1):70.
[6]吕美妮,玉振明.基于DCT零系数和局部标准差的自动聚焦算法[J].激光技术,2018,42(1):66-71.
[7]王烨茹,冯华君,徐之海,等.基于饱和像素剔除的自动对焦评价函数[J].光学学报,2016,36(12):1210001.
[8]张劲峰,张继业.基于函数变化率的变阈值变步长自动聚焦方法[J].计算机工程,2016,42(7):216-219.
[2]蒋涛,左昉,王灵国.严重离焦状态下的自动聚焦实现[J].激光技术,2015,39(6):794-797.
[3]Lyu D,Xia H,Wang C. Research on the Effect of Image Size on Real-Time Performance of Robot Vision Positioning[J]. EURASIP Journal on Image and Video Processing,2018(1):112.
[4]罗钧,孙力,闵志盛.变焦跟踪曲线在对焦中的应用[J].光学精密工程,2011,19(10):2326-2332.
[5]Liu S,Liu M,Yang Z. An Image Auto-Focusing Algorithm for Industrial Image Measurement[J]. EURASIP Journal on Advances in Signal Processing,2016(1):70.
[6]吕美妮,玉振明.基于DCT零系数和局部标准差的自动聚焦算法[J].激光技术,2018,42(1):66-71.
[7]王烨茹,冯华君,徐之海,等.基于饱和像素剔除的自动对焦评价函数[J].光学学报,2016,36(12):1210001.
[8]张劲峰,张继业.基于函数变化率的变阈值变步长自动聚焦方法[J].计算机工程,2016,42(7):216-219.