Comparative Analysis of IR-IR Image Matching Applying the Deep Learning-Based Template Matching Techniques

Template matching technology is a technique that can be applied to various computer vision technologies such as object detection, object recognition, and object tracking. A powerful template matching technique can improve the accuracy and effectiveness of computer vision techniques. In this paper, the template matching techniques are compared with Infrared Radiation images. The techniques are based on deep learning and this paper is to evaluate the performance of deep learning-based techniques for IR images. The techniques consist of QATM(Quality Aware Template Matching), Deep-DIM (Divisive Input Modulation) and SiamFC. The input IR image pairs consist of a search image and a template image, and the template image is an image that crop out of the search image. Each technique is compared and analyzed for runtime and accuracy. The runtime means whole matching time include making feature maps and the accuracy is calculated with RMSE (Root Mean Squared Error).