Fig. 1. 2D image normalization process based on turntable assistance

Fig. 2. Eight neighborhood distribution of key points

Fig. 3. Feature point matching of 2D image

Fig. 4. Feature point description of point clouds

Fig. 5. 3D scanning platform and software system

Fig. 6. Distance error of adjacent point pairs on calibration plate

Fig. 7. Normalization and preprocessing of 2D images. (a) (b) 2D graphs with different perspectives; (c) image after translational rotation transformation for mapped spatial point clouds of Fig. 7(a); (d) 2D image obtained by perspective projection transformation of Fig. 7(c); (e) (f) preprocessed images

Fig. 8. Extraction of feature points. (a) Algorithm of this paper; (b) SIFT algorithm

Fig. 9. Matching of feature points. (a) SIFT; (b) ASIFT; (c) normalization+SIFT; (d) algorithm of this paper

Fig. 10. Collaborative matching results of dual-dimensional feature points. (a) Space posture 1; (b) space posture 2

Fig. 11. Sculpture model splicing. (a) 2D matching; (b) 3D matching; (c) coarse splicing; (d) fine splicing

Fig. 12. Iterations-error curves of traditional and improved ICP algorithms under different noise. (a) Noise of 0.1dB,improved ICP algorithm;(b) noise of 0.5dB, improved ICP algorithm; (c) noise of 0.1dB, traditional ICP algorithm; (d) noise of 0.5dB, traditional ICP algorithm

Fig. 13. Results of partial and integral fine splicing. (a)--(d) Partial fine splicing; (e) (f) overall fine splicing