Multi-Level Median Threshold Bitmaps for HDR Ghost Pixel Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current HDR image creation methods using one-level median threshold bitmaps are insufficient for detecting all ghost pixels, resulting in many missed detections due to their binary nature, which is inadequate for scenes with large luminance differences.
Innovation Solution
The method employs multi-level median threshold bitmaps to generate ghost maps by dividing pixel value ranges into multiple levels, allowing for more accurate detection of ghost pixels across multiple images, and combines these maps using pixel-wise OR to create a combined ghost map for robust ghost detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If one-level median threshold bitmap is used for ghost detection, then the detection process is simple and fast, but many ghost pixels are missed due to insufficient detection accuracy
Solution Approach 1:
The patent segments the detection process into multiple levels by dividing the pixel value range into multiple intervals. Instead of using a single threshold, the method creates multiple median threshold bitmaps at different levels, each handling a specific subset of pixel values. This segmentation allows comprehensive detection of ghost pixels across different luminance ranges while maintaining computational efficiency at each individual level.
Solution Approach 2:
The patent transitions from a one-dimensional binary threshold approach to a multi-dimensional hierarchical threshold structure. By adding the dimension of multiple threshold levels, the system can detect ghost pixels that were previously missed in binary comparisons. Each level adds a new dimension of analysis, enabling detection across the full dynamic range of pixel values.
2Measurement precision
If multi-level median threshold bitmaps are used, then ghost pixel detection accuracy is improved, but computational time and processing complexity increase
Solution Approach 1:
The computation is segmented across multiple levels, where each level processes a specific subset of pixel values. This allows parallel processing and efficient memory access patterns, reducing the overall computational burden despite the increased number of operations. Each level can be processed independently, enabling optimization at each stage.
Solution Approach 2:
The method performs preliminary actions by pre-calculating median thresholds at different levels and preparing multiple threshold bitmaps in advance. This preliminary processing enables faster ghost detection during the actual HDR image creation process, as the threshold structures are ready for immediate use without requiring complex real-time computations.
Data Source
AI summary
A ghost detection method for high-dynamic range (HDR) image creation using multi-level median threshold bitmapping. At each of multiple levels, median threshold bitmaps (MTBs) are generated for each original low-dynamic range image. At each level, the MTBs of the multiple original images are compared to generate a ghost map for each original image which indicates the locations of ghost pixels in the original image. For each original image, the ghost maps generated at the multiple levels are combined to generate a combined ghost map. The multiple combined ghost maps can then be used in creating the HDR image, for example, by discarding pixels in an original image that are indicated by the corresponding combined ghost map as being a ghost pixel.