基于品质因数的边缘检测算子性能优劣客观评价研究

Study on the objective evaluation of the performance of edge detection operator based on the quality factor

数字图像边缘检测算子性能优劣大多基于主观的,非量化的视觉评价,此方法具有人为因素的不确定性,缺乏有效说服力。针对此问题,提出一种边缘检测算子性能优劣评价的客观量化指标,即选用Canny算子作为优劣评价的参考标准,使用品质因数R作为客观评价的量化指标,使用距离最近原则下的查表搜索方法实现品质因数R的计算。通过人造图像和自然图像的仿真实验,实验量化结果及主客观评价的一致性说明此指标的有效性和此方法的可行性,为边缘检测算子的选择提供参考依据,为算子性能改进提升提供一种量化评价指标。

In digital image processing, it is commonly using the non-quantifiable and subjective evaluation for the edge detector’ s performance, so that method is lack of convincingness for its uncertainty because of human factors. Aim at this, here a qualified ob-jective evaluation index is proposed:choosing the Canny operator as the criterion, quality factor R as the qualified objective evalua-tion index, in order to calculate the quality factor R, a search method under the principle of shortest distance is used. Via the experi-ments of obtaining this index from the artificial and nature of digital images respectively, the numerical result of experiments and the coherence of the subjective and objective evaluations indicate the index is valid and this method is viable. This index can be a refer-ence of selecting one of the edge detect operators, also it can be a valid index during improving or promoting the edge detector’s per-formance.