Fig. 1. The H channel controls the range of different color values

Fig. 2. Flowchart of the proposed algorithm

Fig. 3. Comparison of segmentation effects. (a) Original sintering images; (b) K=2; (c) K=3; (d) K=4

Fig. 4. Comparison of color extraction effect. (a) Image segmented by K-mean algorithm; (b) red fire area after color extraction

Fig. 5. Geometric feature data. (a) Area; (b) long axis; (c) minor axis; (d) eccentricity; (e) Euler number; (f) aspect ratio

Fig. 6. Classification results of FCM algorithm. (a) Before denoising; (b) after denoising

Fig. 7. Comparison of number of iterations. (a) K-FCM; (b) conventional FCM

Fig. 8. Comparison of classification membership. (a) K-FCM; (b) conventional FCM

Fig. 9. Comparison of clustering results. (a) K-FCM; (b) conventional FCM