Transform Domain Image Merging for HDR Generation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing imaging devices struggle to capture high dynamic range (HDR) images due to the high cost of image sensors and the complexity of processing multiple images with different exposure settings, which requires expensive equipment and computation-intensive algorithms.
Innovation Solution
The method involves merging multiple non-HDR images directly in a transform domain, such as the DCT or DFT domain, without decompression or secondary compression, using energy values from coefficient blocks to determine pixel luminance levels and combine images to create an HDR image, thereby reducing computational complexity and cost.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If multiple non-HDR images are merged in the spatial domain using existing approaches, then HDR image quality can be achieved, but the processing complexity and computational cost increase significantly
Solution Approach 1:
The patent transforms the image merging operation from the spatial domain to the frequency domain using Fast Fourier Transform (FFT). By converting images to frequency domain representations and performing merging operations there, the system achieves HDR image quality while reducing spatial processing complexity. The frequency domain operations allow for more efficient combination of multiple exposed images without requiring complex pixel-by-pixel processing in the spatial domain.
2Adaptability or versatility
If multiple images with different exposure settings are processed to create HDR images, then dynamic range is improved, but the computation time and processing power requirements increase
Solution Approach 1:
The system performs FFT transformation to move image data to the frequency domain, where merging operations can be performed more efficiently. This dimensional change enables the system to process multiple images with different exposure settings to achieve extended dynamic range while reducing the computational time and processing power requirements compared to traditional spatial domain methods.
Solution Approach 2:
The patent replaces complex spatial domain merging algorithms with frequency domain operations. By substituting the mechanical/spatial processing approach with a frequency-based mathematical transformation, the system achieves the same HDR effect with reduced computational complexity and faster processing speeds.
3Measurement precision
If image data is decompressed and transformed from transform domain to spatial domain for merging, then accurate luminance level determination is achieved, but the processing steps and costs increase
Solution Approach 1:
Instead of decompressing and transforming images to the spatial domain before merging (the conventional approach), the patent inverts the workflow by performing merging operations directly in the frequency domain. This inversion eliminates the need for decompression and transformation steps while maintaining the ability to accurately determine luminance levels through frequency domain coefficient analysis.
Data Source
AI summary
Techniques are provided to generate high or wide dynamic range image from two or more input images of different exposure settings by directly merging coefficients derived from the input images in a transform domain. Energy values may be determined from coefficients blocks derived from the input images. The energy values may be compared with thresholds to determine weight factors for the coefficient blocks. An output coefficient block in the transform domain, used in or used to generate the output image, may be determined as a weighted combination of the coefficient blocks in the transform domain derived from the input images. If input images are compressed in transform domain, an output image can be generated without performing decompression in transform domain.