Time of Flight Depth Image Noise Reduction via Phase Filtering
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Time of Flight (ToF) methods for acquiring depth images often result in noisy depth images, which existing technologies have not effectively addressed.
Innovation Solution
A method and apparatus that receive and process multiple phase images with different phases, using filters such as Gaussian, average, median, or bilateral filters to eliminate noise, and acquire a depth image by processing these filtered images, including sequentially irradiating a subject with beams of different phases and detecting reflection beams to generate phase images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If ToF method is used to acquire depth image in real-time, then depth image acquisition speed is improved, but noise in depth image increases
Solution Approach 1:
The patent applies preliminary action by performing noise elimination on phase images before acquiring the final depth image. Multiple phase images are pre-processed using filtering operations (Gaussian, median, bilateral filters) to remove noise artifacts, and only then is the depth image calculated from the cleaned phase images, preventing noise propagation to the final result
Solution Approach 2:
The patent introduces an intermediary processing step using multiple phase images with different phases (0°, 90°, 180°, 270°) as mediators between the raw ToF measurement and the final depth image. These intermediate phase images serve as a buffer to eliminate noise through filtering operations before the final depth calculation, resolving the contradiction between speed and quality
2Measurement precision
If multiple phase images are processed to eliminate noise, then depth image quality is improved, but processing time increases
Solution Approach 1:
The patent applies partial action by selectively processing only the necessary phase images for noise elimination. Instead of processing all possible phase data, the method focuses on eliminating noise from key phase images (particularly those with 180° phase difference) to achieve sufficient noise reduction without excessive processing time, balancing quality improvement with time constraints
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The method effectively reduces noise in depth images, improving their accuracy and quality by utilizing noise elimination techniques on phase images before acquiring the depth image, as demonstrated through simulation results.
Implementation Method 1
an image sensor acquiring the plurality of phase images by detecting a plurality of reflection beams which are reflected from the subject
Implementation Method 2
a ToF (time of flight) method utilizes a return time of an infrared ray after irradiating a subject with the infrared ray
Implementation Method 3
The eliminating of the noise from the plurality of phase images may include utilizing at least one of a Gaussian filter, an average filter, a median filter, and a bilateral filter
Data Source
AI summary
A method of acquiring a depth image and an image acquiring apparatus. The method of acquiring the depth image includes receiving a plurality of phase images with respect to a subject, the plurality of phase images having phases different from one another, eliminating noise from the plurality of phase images, and acquiring the depth image with respect to the subject utilizing the plurality of phase images from which noise has been eliminated.