Fig. 1. Flow chart of the proposed algorithm in this paper

Fig. 2. Detection of contour feature points

Fig. 3. Featured contour. (a), (d) Contours of thermal infrared image and visible light image extracted by traditional method; (b), (e) Polygon approximation of contours in reference [10]; (c), (f) Featured contours extracted by proposed FCQ algorithm

Fig. 4. Feature contours of thermal infrared and visible images extracted under complex background

Fig. 4. [in Chinese]

Fig. 5. Algorithm flow chart for generating featured contour quadrilateral

Fig. 6. Calculation of displacement variable deviation value

Fig. 7. The first registration result of thermal infrared image and visible image in experiment 1

Fig. 8. The second group of thermal infrared image and visible image registration results in experiment 1. (a), (b) Thermal infrared image and its featured contour quadrilateral; (c), (d) Visible image and its featured contour quadrilateral; (e), (f) Registration graph and fusion graph of the FCQ algorithm using improved homography transformation; (g), (h) Registration graph and the fusion graph of FCQ algorithm applying traditional homography transformation

Fig. 8. [in Chinese]

Fig. 8. [in Chinese]

Fig. 9. Registration results of thermal infrared image and visible image in experiment 2

Fig. 10. Registration result of thermal infrared image and visible image in experiment 3. (a) Thermal infrared image; (b) Visible image; (c) Registration image; (d) Fusion image; (e),(f) Registration graph and the fusion graph without applying the IHT

Fig. 10. [in Chinese]

Table 1. Registration matrix parameters in experiments

Table 2. MSE, RMSE of the three experiments and the comparison groups