Fig. 1. Workflow of algorithm

Fig. 2. Schematic of RPCA scene sample training. (a) One of the scene samples; (b) target sample; (c) training process

Fig. 3. Schematic diagram of RPCA in image segmentation. (a) Image to be segmented; (b) high-rank noises (targets)

Fig. 4. Effect of morphology operation. (a) Adaptive thresholding segmentation to high-rank noises; (b) morphology operation; (c) corresponding pixels in original image

Fig. 5. Improved k-means clustering process

Fig. 6. Schematic of SVM classification. (a) Input regions; (b) output targets with label 1; (c) output targets with label 0; (d) training and classification progress

Fig. 7. Schematic of epipolar line rectification

Fig. 8. Template searching matching point through epipolar line

Fig. 9. Positioning of right image objects. (a) Targets labeled 1; (b) targets labeled 0

Fig. 10. Relationship between new and old coordinates

Fig. 11. Illustration of advanced pass through denoising

Fig. 12. Schematic of radium filtering

Fig. 13. Contrast before and after denoising. (a) Original point clouds; (b) denoised point clouds

Fig. 14. Experimental platform

Fig. 15. Schematic of location effect of different algorithms in complex environment. (a)-(d) Messy background with identical color; (e) stacked targets with weak light; (f) messy background with nonlinear light

Fig. 16. Data of algorithm efficiency. (a) Contrast of calculation quantity; (b) contrast of running time

Fig. 17. Experimental data of algorithm error. (a) Cylinder measuring error; (b) square measuring error; (c) gear measuring error