A robust sub-pixel subdivision algorithm for image-type angular displacement measurement

Highlight
- For angle measurement, we established a robust subdivision algorithm which can minimize the influence of image noise and lens defocus effectively.
- We do lots of simulation between proposed algorithm and traditional algorithm to analyze the performance of proposed algorithm.
- We applied the algorithm in an angle sensor and achieved 221 resolution and 12.85″ precision when the grating diameter was 38mm. This result is better than traditional.
- We are researching a new angle measurement by using image detector and this is a newest issues. There are no researches about the subdivision algorithm of image type angle measurement. So we propose a new subdivision algorithm which is proved that is more suited to image type angle measurement.

The use of an image detector to receive grating images and measure angle displacement via image processing is a relatively new technique, which yields higher resolution and better precision than the traditional moiré fringe method. To improve the robustness of image-type angle measurement, this paper proposes a robust sub-pixel subdivision algorithm based on the least square method. Firstly, by analyzing the characteristic of grating image, a new subdivision algorithm is established based on the least square method. Secondly, the simulations of robustness are completed to prove the performance in theoretically. Lastly, the proposed algorithm is used in a typical image-type angle sensor to test the performance in real case. By test, the proposed method is shown to be more accurate and with better robust than the traditional algorithm (centroid algorithm). In a typical image-type angle sensor, it successfully achieves a resolution of 0.62″ (21-bit), 213-fold subdivision resolution, and precision of 12.85″. The results presented here may provide a theoretical and technological foundation for further research on small-size, high-resolution photographic rotary encoders.