Abstract—In the study, we examine the problem of diagnosing distortions in the resulting image, the so-called “blind estimation” task. The questions of identification of the type and parameters of linear smoothing homogeneous distorting operators of three types are investigated: circular form with a rectangular profile, circular form with a Gaussian profile, and linear form with a rectangular profile. The identification of distortions is carried out by means of the image spectrum density analysis using the average radial profile and average directional profile. The diagnostic algorithms for each of the considered distortion types are proposed and studied. The influence of noise and data accuracy on the result reliability and the capabilities of diagnostic algorithms in the case of simultaneous superposition of several distortions are demonstrated.

摘要—在这项研究中，我们研究了诊断所得图像失真的问题，即所谓的“盲估计”任务。研究了三种类型的线性平滑齐次失真算子的类型和参数的识别问题：具有矩形轮廓的圆形，具有高斯轮廓的圆形和具有矩形轮廓的线性。通过使用平均径向轮廓和平均方向轮廓的图像光谱密度分析来进行失真的识别。提出并研究了每种考虑的失真类型的诊断算法。演示了噪声和数据精度对结果可靠性和诊断算法功能的影响，这些结果同时叠加了几种失真。