Dual-FOV Image Processing for Neural Color and Dynamic-Range Alignment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing systems struggle to integrate dual-field-of-view (FOV) image streams effectively, particularly in VR and AR applications, due to discrepancies in color, brightness, and dynamic range, which are exacerbated by different processing techniques and lighting conditions, leading to inconsistent and unrealistic user experiences.
Innovation Solution
A method and system that process wide and narrow FOV image streams using a first processing technique, store color, gamma, and semantic information, and utilize a neural network to enhance the narrow FOV stream, followed by blending to ensure color consistency and dynamic range alignment, employing hardware-based ISPs, software-based ISPs, and neural networks for adaptive processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If different processing techniques are used for wide and narrow FOV image streams, then each stream can be optimized for its specific characteristics, but color and brightness consistency between streams deteriorates
Solution Approach 1:
The patent introduces an intermediary processing stage where the first image stream is processed to generate reference information (color data, gamma values, tone mapping parameters) that serves as a mediator to guide the processing of the second image stream. This intermediary reference information ensures that both streams, despite using different processing techniques, converge to consistent visual output characteristics.
Solution Approach 2:
The patent dynamically adjusts processing parameters (color temperature, gamma curves, tone mapping settings) based on the content and characteristics of each image stream. By changing these parameters adaptively while maintaining reference to the first stream's processed characteristics, the system optimizes each stream individually while preserving overall consistency.
2Stability of the object's composition
If manual adjustments are made to align image streams, then color and brightness consistency improves, but processing time and complexity increase
Solution Approach 1:
The patent performs preliminary processing on the first image stream to extract and store reference information (color profiles, gamma values, tone mapping parameters) before processing the second stream. This preliminary action prepares the necessary alignment data in advance, enabling automatic consistency maintenance during real-time processing without requiring manual adjustments for each frame.
Solution Approach 2:
The patent implements a feedback mechanism where the processed first image stream provides reference information that continuously guides the processing of the second stream. This closed-loop feedback system automatically maintains color and brightness consistency dynamically, replacing manual adjustments with real-time adaptive control.
3Productivity
If simplistic blending techniques are used to combine image streams, then processing speed improves, but visual coherence and realism deteriorate
Solution Approach 1:
The patent applies sophisticated parameter-based processing (color temperature adjustment, gamma correction, tone mapping) to each image stream based on its specific characteristics and the reference information from the first stream. These parameter transformations ensure visual coherence and realism in the blended output while maintaining efficient automated processing suitable for real-time applications.
Data Source
AI summary
Disclosed is a method and an apparatus for processing dual-stream images. The method includes processing a wide field-of-view (FOV) image stream using a first processing technique. The method further includes storing one or more of color information, gamma or tone mapping information, semantic information from the wide FOV image stream. The method further includes processing a narrow FOV image stream using a neural network based on the one or more of the color information, the gamma or tone mapping information, and the semantic information. The method further includes blending the processed wide FOV image stream and the processed narrow FOV image stream to generate a final image, for display on a display device.