基于汉语手指语字母识别的手指角度模糊融合

以汉语手指语拼音字母为研究对象,根据自行研制的数据手套CAS_Glove上传感器的特点,采用了弯曲传感器和外展传感器先后分别进行模糊融合的策略。针对弯曲传感器的模糊融合,提出一种分类融合方法,分别采用弱t模算子和弱t共模算子,再对其结果进行转换,最后用MICA算子进行弯曲传感器的最终融合。针对外展传感器融合的特点,使用了弱t模算子。通过和其它模糊融合方法的实验对比,证明该方法在数据手套进行角度融合时,优于其它模糊融合方法。

Fuzzy Fusion of Fingers' Angles Based on the Recognition of Chinese Sign Language Letters

In this paper,the techniques of fuzzy fusion are adopted according to the features of the sensors of a self-developed dataglove-CAS_Glove. The bend sensors and abduction sensors are respectively dealt with based on the features of the letters of Chinese finger language. In relation to the fuzzy fusion of bend sensors,a classified fusion method is presented. In this method,firstly,weak t norm operator and weak t conorm operator are used separately. Then the results are respectively transformed to fit in with MICA operator. Finally,MICA operator is used to get the fusion result of bend sensors. In addition,weak t norm operator is used in the fusion of the abduction sensors. A series of experiments are done to compare this method with other fuzzy fusion methods. Experiments show that this method is superior to other fuzzy methods when making angle fusion of this kind of dataglove.