RGBIR Image Conversion With Subsampling for Lower ISP Cost
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
RGBIR image sensors face challenges in adjusting resolution to meet both high and low-resolution data needs, particularly in smart camera systems, and existing image signal processors struggle to process RGBIR data effectively due to the presence of IR pixels, leading to inefficiencies in memory consumption and computation costs.
Innovation Solution
An RGBIR image processing system that includes an RGBIR-Bayer converter and a Bayer-RGB converter, utilizing a subsampler to decimate RGBIR data and a color adjusting device to generate compatible RGB color images, reducing data size and computational costs through odd-numbered subsampling and color processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If RGBIR raw data is processed using conventional ISP pipeline, then color image can be generated, but memory consumption and computation cost increase significantly
Solution Approach 1:
The patent extracts only the necessary processing steps for RGBIR data and removes unnecessary conventional ISP pipeline stages. The simplified pipeline directly converts RGBIR raw data to color image data without performing full demosaicing and color correction, significantly reducing computation cost while maintaining essential image processing capability.
Solution Approach 2:
The patent segments the image processing pipeline into distinct stages: RGBIR raw data conversion, optional subsampling, and color image generation. This segmentation allows selective application of processing steps based on resource constraints, enabling the system to balance between processing capability and computation cost.
2Measurement precision
If RGBIR sensor captures high resolution data, then image quality improves, but memory consumption increases
Solution Approach 1:
The patent applies partial action by performing subsampling on RGBIR raw data before processing. Instead of processing all high-resolution pixels, the system selectively processes a subset of pixels (e.g., every other pixel), reducing memory consumption while maintaining sufficient image quality for analytics purposes.
Solution Approach 2:
The patent changes the resolution parameter of the input data by applying subsampling. This parameter change reduces the data size and memory requirements while preserving the essential image information needed for smart camera analytics, effectively balancing image quality and memory consumption.
3Adaptability or versatility
If full RGBIR processing pipeline is implemented, then complete color image processing is achieved, but device complexity increases
Solution Approach 1:
The patent creates a universal processing approach that handles RGBIR data through a simplified pipeline that can be adapted to different smart camera configurations. The same basic conversion and subsampling steps work across different resolutions and applications, reducing device complexity while maintaining processing compatibility.
Data Source
AI summary
A red, green, blue and infrared (RGBIR) image processing system includes an RGBIR-Bayer converter that converts RGBIR raw data to Bayer raw data, and a Bayer-RGB converter that converts the Bayer raw data to RGB color data. The Bayer-RGB converter includes a color adjusting device that adjusts color information of the Bayer raw data, thereby generating adjusted data.