Multi-frequency phase unwrapping method and device

Abstract

The present invention discloses a multi-frequency phase unwrapping method and device, belonging to the technical field of computer vision. The method comprises the steps of: employing phase measurement values of a plurality of modulation signals in different modulation frequencies to perform preliminary phase unwrapping to obtain multi-group unwrapping coefficient constraints; based on the Chineseremainder theorem, expanding the multi-group unwrapping coefficient constraints to obtain expansion phase measurement value; obtaining a pseudo-distance estimation value through the phase measurementvalues expanded by a weighed even value combination; based on kernel density estimation of the pseudo-distance estimation value, obtaining an unpacking probability; constructing and obtaining a predicted hypothesis subset through the unpacking probability; calculating probable values of predication phase noise and predication phases based on the predicted hypothesis subset; and employing the kernel density estimation to perform ranking of the probable values of the predication phases and determine an unwrapping phase value. The multi-frequency phase unwrapping method and device are suitable for depth decoding of a Kinect v2 sensor, can increase the coverage range of effective pixels, and can maintain the timeliness.

Description

technical field [0001] The invention relates to a multi-frequency phase unwrapping method and device, belonging to the technical field of computer vision. Background technique [0002] In the time-of-flight sensor, the modulated light signal is irradiated on the object in the environment, and then the reflected light signal is captured in the pixel array of the sensor, and the distance between the object in the environment and the sensor is calculated by the time difference of light traveling in the air. The distance is the depth pixel value, and the purpose of phase unwrapping is to select the correct period of the optical modulation signal (for i.e. the solution to determine k), because a wrong period can lead to a large depth error. [0003] The classic solution to the multi-frequency unwrapping problem is to use the Chinese remainder theorem, which is fast but implicitly assumes noise-free data, by instead generating multiple unwrappings for each frequency, and then fo...

AI Technical Summary

Problems solved by technology

However, the performance of the actual data is not shown

After the existing technology decodes the depth, because the effective range of the measured depth value is relatively small, the coverage of effective pixels is also relatively small, and the performance of the actual data of the pixels is not displayed.

Images