Fig. 1. Images with different blur levels and their corresponding edge images. (a)(b) Primitive clear image and corresponding edge image; (c)(d) once fuzzy image and the corresponding edge image; (e)(f) secondary fuzzy image and corresponding edge image

Fig. 2. Edge difference images of images with different blur levels. (a) Primitive clear image and once fuzzy image edge differences; (b) once fuzzy image and secondary fuzzy image edge differences

Fig. 3. Calculation flowchart of no reference image blur parameter

Fig. 4. Schematic representation of GMS algorithm

Fig. 5. Two cell models. (a) 8 neighborhood; (b) X neighborhood

Fig. 6. Original clear image samples and image samples with added blur noise. (a) Original clear image sample in 00 sequence; (b) image sample after adding blur noise

Fig. 7. Two algorithms for blurring image classification results on the 00 sequence. (a) ISVD classification results; (b) reBlur classification results

Fig. 8. Original image pairs and matching results of different algorithms. (a) Initial image pair; (b) ORB matching results; (c) GMS matching results; (d) Matching results of proposed algorithm

Fig. 9. Results of using reBlur to remove blurred images in each sequence. (a) 00 sequence; (b) 01 sequence; (c) 02 sequence; (d) 03 sequence; (e) 04 sequence; (f) 05 sequence

Fig. 10. Different sequence trajectories and real trajectories obtained by different methods comparison diagram. (a) 00 sequence; (b) 03 sequence