基于熵权的图像分割性能灰色评判

Entropy Weight-based Grey Evaluation of Performance of Image Segmentation

为了克服传统的灰色综合评判中主观赋权的局限性,借助于信息工程学中的熵的概念,根据各评价指标的差异程度来修正权重,建立了一种新的综合评判模型——熵权灰色综合评判模型,并将该模型应用到图像分割性能评判当中。讨论影响图像分割性能的各种因素,从评判的灰色性入手,运用灰色数学的原理和方法对各因素从理论上进行定量分析。对因素权重的确定,不是单凭主观判断,而是采用熵权系数法进行客观计算。评价对象的固有信息得到充分利用。避免了以往评价中只强调过程的某几项少数指标而忽略其它指标的缺点。因此,该模型具有更好的有效性和实用性。

In order to overcome the subjective limitation in traditional grey comprehensive evaluation, a new model for comprehensive evaluation, i.e. entropy weight comprehensive evaluation model, was established by using the concept of entropy in information engineering science and the modified weight value with differential degree of evaluation indexes. And the model was applied to evaluating the performance of image segmentation. All kinds of factors having influence on image segmentation performance were analyzed theoretically and quantitatively by the principle and the method of grey mathematics. The weight of the factor was calculated objectively by an entropy weight-based coefficient method instead of only subjective judgement. And the inherent information of the evaluated subject is used fully in evaluation so that the drawback of the traditional evaluation can be avoided, in which only a few indicators of process are emphasized while others are out of consideration. Thus the model is more effective and feasible.