Image Signal Processor HDR Stitch Mapping for Fast Dynamic Range
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Image processing systems, such as cameras, struggle to capture high dynamic range (HDR) images due to the limited brightness sensing capabilities of image sensors, resulting in scenes captured in backlit conditions appearing differently from how human eyes perceive them, with backgrounds overly bright and subjects too dark.
Innovation Solution
An image signal processor that generates HDR images by reducing the resolution of multiple images to create low-level images, then uses a saturation map and motion map to synthesize a stitch map, applying weights to generate an HDR image with improved dynamic range and signal-to-noise ratio.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Illumination intensity
If multiple images with different exposure times are captured to increase dynamic range, then the dynamic range of the image is improved, but the processing complexity and time required to generate the HDR image increases
Solution Approach 1:
The patent segments the HDR image generation process into distinct functional modules: a low level image generation circuit that reduces resolution of input images, and an HDR circuit that performs map generation (saturation map, motion map, stitch map) and weighted synthesis. This segmentation allows complex HDR processing to be divided into manageable stages, improving processing efficiency while maintaining enhanced dynamic range.
2Illumination intensity
If multiple images with different exposure times are captured to increase dynamic range, then the dynamic range of the image is improved, but the processing time required to generate the HDR image increases
Solution Approach 1:
The patent applies preliminary action by generating low level images with reduced resolution before performing the computationally intensive HDR processing. By pre-processing the input images to reduce their resolution, the subsequent generation of saturation maps, motion maps, and stitch maps requires less processing time, thereby reducing overall HDR generation time while still achieving enhanced dynamic range in the final image.
3Productivity
If the resolution of images is reduced to create low level images, then the processing speed is improved, but the image quality and detail are degraded
Solution Approach 1:
The patent applies local quality by using reduced resolution low level images specifically for generating motion maps and stitch maps, which require processing speed rather than fine detail. The original full-resolution images are preserved and used for the final HDR synthesis, ensuring that image quality is maintained in the regions where detail matters most while achieving fast processing for motion analysis.
4Measurement precision
If saturation map and motion map are synthesized to create stitch map, then the HDR image accuracy is improved, but the processing complexity increases
Solution Approach 1:
The patent introduces the stitch map as an intermediary element that bridges the saturation map and motion map. The stitch map is generated by synthesizing these two maps and serves as a weighting guide for the final HDR image generation. This intermediary approach improves HDR accuracy by systematically combining saturation and motion information, while the modular structure keeps processing complexity manageable through clear separation of functions.
Data Source
AI summary
An image signal processor, a method of operating the image signal processor, and an application processor including the image signal processor are provided. The image signal processor includes: a low level image generation circuit configured to generate a plurality of low level images by reducing resolutions of a plurality of images; and a high dynamic range (HDR) circuit configured to generate a saturation map and a motion map based on the plurality of low level images, generate a stitch map by synthesizing the saturation map with the motion map, and generate an HDR image by applying a weight to the generated stitch map.