摘要
针对现有算法只能进行单一目标模糊参数的鉴别的缺点,提出利用局部标准差和现有算法相结合的方法同时鉴定多目标的模糊参数。通过对现有算法的分析,对模糊图像进行局部标准差滤波,提取标准差较大的模糊边缘图像块,对各图像块进行筛选、归类,然后应用Radon变换及改进倒谱法分别对每个图像块进行运动模糊参数的鉴别,再通过加权平均处理,进一步精确参数结果。算法对鉴别多目标模糊参数有良好的精度,而且提高了运动模糊参数鉴别的稳定性和准确性。
The existing algorithms can only identify blur parameters for a single objective. To overcome this shortcoming, this paper proposes to use the method combining the local standard deviation and the existing algorithms tosimultaneously identify the multiobjective motionblur parameters. First the motionblur image was filtered by localstandard deviation to extract the larger standard deviation blur edge image blocks for screening and classification.Then, motion blur parameters of every image blocks were obtained based on Radon transform and the improved cepstrum algorithm. And then, the results of more accurate parameters were further retrieved via weighted average processing. The experimental results show that this algorithm of multiobjective blur parameters identification has highaccuracy and can enhance the stability and accuracy of identification of motion blur parameters.
出处
《应用科技》
CAS
2016年第3期17-22,共6页
Applied Science and Technology
基金
国家自然科学基金项目(61371175)
关键词
多目标
参数估计
运动模糊
局部标准差
图像分块
加权平均
multi-objective
parameter estimation
motion-blur
local standard deviation
image-block
weighted average method