Multi-scale Consistency Map for HDR Ghost Detection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing HDR image creation techniques are sensitive to noise and fail to detect certain types of differences between images, particularly when small windows are used, leading to incomplete ghost artifact detection, while large windows may miss small ghost-inducing regions.
Innovation Solution
A method that generates individual consistency maps using moving windows of multiple sizes, combines these maps, and binarizes them to create a final consistency map, which is then used to generate ghost-weight maps to identify and address ghost-inducing regions in HDR image creation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If small fixed-size windows are used for ghost detection, then the detection algorithm is computationally efficient and localized, but it becomes sensitive to noise and fails to detect certain types of differences between brackets
Solution Approach 1:
The patent divides the detection process into multiple segments by using windows of different sizes (e.g., 7x7, 15x15, 31x31 pixels) to analyze the same image region. Each window size segment captures different types of changes: small windows detect fine-grained noise and local artifacts, while large windows detect broader scene changes. The consistency maps from different window sizes are then combined to achieve comprehensive ghost detection that overcomes the limitations of any single window size.
2Area of stationary object
If large windows are used for ghost detection, then small ghost-inducing regions can be detected, but small ghost-inducing regions may be missed
Solution Approach 1:
The patent adds the dimension of window size variability to the detection process. Instead of using a single fixed window size, the system varies the window size dimension to create multiple consistency maps at different scales. This multi-scale approach allows the detection system to capture ghost artifacts at various spatial scales simultaneously, ensuring that both small and large ghost-inducing regions are detected with appropriate precision.
3Productivity
If single window size is used for comparison, then the processing is simple and fast, but certain types of differences between brackets are not detected
Solution Approach 1:
The patent changes the parameter of window size to improve detection precision. By systematically varying the window size parameter and generating consistency maps at multiple scales, the system achieves comprehensive detection of different types of bracket differences. The final ghost probability map combines results from all window sizes, ensuring that no type of ghost artifact is missed while maintaining processing efficiency through the structured multi-scale approach.
Data Source
AI summary
A method for detecting ghost artifact in multiple images during high dynamic range (HDR image creation. For each pair of images, multiple individual consistency maps are generated by calculating a consistency function using moving windows of different sizes. The individual consistency maps are combined into a combined consistency map and then binarized. When combining the multiple individual consistency maps, they are first binarized using predetermined threshold values that are window-size-specific. The threshold values are developed beforehand using training images and machine leaning. The final binarized consistency map indicates whether the pair of images are consistent with each other at each pixel location. Then, a ghost-weight map is generated for each image based on the multiple final consistency maps. The ghost-weight map indicates the likelihood of each image pixel being ghost-inducing. The HDR image is generated using the set of images and the corresponding ghost-weight maps.