IR Texture Removal in Stereoscopic Depth Sensors
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current techniques for obtaining high-quality images that include and exclude infrared (IR) textured patterns in stereoscopic depth sensing are costly, power-intensive, and increase device size, while also limiting frame rates and being susceptible to motion artifacts.
Innovation Solution
A system that applies a color correction transform to raw image data from an image sensor with a color filter array to remove or retain IR texture patterns, using a dual pipeline image signal processor to generate images with reduced or no IR texture residuals, allowing for efficient and effective IR texture pattern management.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If separate imagers are used to obtain color images without IR texture in addition to images with IR texture, then image quality is improved, but device size and cost increase
Solution Approach 1:
The patent divides the image processing task into two separate processing pipelines within a single imager: one pipeline processes images with IR texture for depth sensing, while the other pipeline removes IR texture for color imaging. This segmentation of processing functions eliminates the need for separate physical imagers, reducing device size while maintaining image quality for both purposes.
2Measurement precision
If separate imagers are used to obtain color images without IR texture, then image quality is improved, but power consumption increases
Solution Approach 1:
The patent merges the functionality of multiple imagers into a single imager by implementing different image processing pipelines that operate on the same captured data. The single imager captures both IR-textured and non-IR-textured images, and the processing pipelines separately handle depth sensing and color imaging needs, thereby reducing power consumption while maintaining image quality.
3Area of stationary object
If time multiplexing techniques are used to obtain images with and without IR texture, then device size is reduced, but frame rate decreases
Solution Approach 1:
The patent enables continuous capture of both IR-textured and non-IR-textured images simultaneously through a single imager with multiple processing pipelines. This continuous parallel processing eliminates the frame rate limitations inherent in time multiplexing approaches, where images must be captured sequentially, thereby maintaining high frame rates while keeping device size small.
4Area of stationary object
If time multiplexing techniques are used to obtain images with and without IR texture, then device size is reduced, but power consumption increases
Solution Approach 1:
The patent implements self-service processing within a single imager where the captured image data is simultaneously processed through multiple pipelines: one for depth sensing with IR texture and another for color imaging without IR texture. This self-service approach eliminates the need for separate imagers and their associated power consumption, while maintaining the ability to produce both types of images efficiently.
5Area of stationary object
If image sensors with modified Bayer pattern are used to extract RGB image data, then device size is reduced, but image quality deteriorates
Solution Approach 1:
The patent uses a standard Bayer pattern color filter array in the single imager, maintaining normal color sensitivity parameters. By processing the captured images through specialized pipelines that remove or retain IR texture as needed, the system achieves both size reduction and maintains high image quality, avoiding the quality deterioration that would result from modifying the Bayer pattern itself.
Data Source
AI summary
Systems, devices, and techniques related to removing infrared texture patterns used for depth sensors are discussed. Such techniques may include applying a color correction transform to raw input image data including a residual infrared texture pattern to generate output image data such that the output image data has a reduced IR texture pattern residual with respect to the raw input image data.