A thermal face recognition under disguised conditions using model fusion is proposed. The proposed model fusion has three main approaches: Linear Support Vector Machine Classifier (Linear SVC), Convolutional Neural Network (CNN) and Ordinary Least Square (OLS). A grid of 22 thermal face points based on physiological information is extracted for training Linear SVC. The support vectors of testing images using Linear SVC are calculated to find the hyperplane for classification. The novelty is that we firstly apply temperature information in face recognition. In CNN model, the multiple layers of convolutional layer are utilized to extract more effective features. The fully connected layer (FC layer) is trained using the feature matrix of the last convolutional layer. This FC layer is then a classifier to identify the category of the test image. In the training phase, the predicted values from above two approaches are provided to the OLS for linear regression. The OLS assigns weighting values to these two approaches. This can effectively compensate for the advantages and disadvantages of two approaches. In addition to the comparison with the traditional thermal face recognition, an experiment under disguised conditions was conducted. Experimental results of the proposed method outperform the existing methods.