Hossein Boroumand Noghabi - department of Computer Engineering Islamic Azad University of Zanjan Zanjan, Iran
Fatemeh Abdollahi - Department of Computer Engineering Islamic Azad University of Zanjan Zanjan, Iran

Thermography of infrared as science and access to the data analysis of thermal imaging machines, non-contact. Capability of infrared defects in the early detection equipment and industrial processes faulty and reduce the performance of the system, which maintenance and it would be catastrophic. using image processing tool on thermal images of the possibility of the failure to detect automatically provides that such a technique to thermal infrared images are currently in the fields of medicine and non - destructive use of existing approaches, has a lot of time , speed a little identify defects are industrial equipment to tackle this problem to improve Accuracy, Sensitivity, Precision , specificity , in this paper, a new strategy for the automatic recognition of flaws in the thermal images using TWSVM algorithm proposed. In the present study of statistical feature based on first order, GLCM for features extraction and TWSVM classification for thermal conditions, according to the standard of NETA has been used. Based on the results of the study shows that the automatic recognition system based on the taxonomy TWSVM has Accuracy 94.7 % , Sensitivity 95.2 , Precision 96.7 %, specificity 93.7 % compared with the SVM classification diagnosis and treatment time reduction capability increases system

TWSVM, thermography, mechanical equipment, image processing