Image Stitching Weighting Arrays for Faster Low-Memory Blending
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image combining methods require high-capacity memory for storing weighting factors, especially when high-quality images are combined, and may suffer from increased memory requirements and slow processing times.
Innovation Solution
An image combining method that determines a weighting direction based on the size of overlapping regions in images, using one-dimensional weighting factor arrays to reduce memory usage and enable high-speed image combining, while avoiding calculation errors.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If weighting factors for all overlapping regions are determined in advance and stored in memory, then image combining speed is improved, but memory capacity requirements increase
Solution Approach 1:
The patent segments the weighting factor storage into two parts: a small reference table containing weighting factor patterns, and runtime memory that stores only the minimum necessary weighting factors for current overlapping regions. This segmentation reduces overall memory requirements while maintaining fast access to weighting factors during image combining operations.
Solution Approach 2:
The patent transforms the traditional two-dimensional weighting factor array into a one-dimensional array by mapping 2D coordinates to 1D indices using the formula: index = y * width + x. This dimensional transformation reduces memory overhead and simplifies memory access patterns, improving both speed and memory efficiency.
2Manufacturing precision
If the amount of data of weighting factor is increased to improve image quality, then image quality is improved, but memory capacity requirements further increase
Solution Approach 1:
The patent changes the data representation format of weighting factors from high-precision floating-point numbers to lower-precision fixed-point or integer representations. This parameter change maintains sufficient image quality while dramatically reducing the memory capacity required to store weighting factors, allowing higher precision to be used only where absolutely necessary.
3Manufacturing precision
If many images are combined to achieve high resolution or wide range imaging, then image quality is improved, but processing time increases
Solution Approach 1:
The patent performs preliminary actions by pre-calculating and storing weighting factor patterns in a reference table during system initialization. This preliminary action eliminates the need for real-time weighting factor calculations during image combining, significantly reducing processing time when combining multiple high-resolution images while maintaining image quality.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
An image combining method includes a step of preparing picked-up images each provided with an overlapping region, a step of determining at least one of X and Y directions as a weighting direction for each overlapping region, a step of setting a one-dimensional weighting factor array in which a weighting factor gradually decreases toward an outer side of each picked-up image in the weighting direction in the overlapping region, a step of obtaining a weighted pixel value of each overlapping pixel included in the overlapping region in each picked-up image by weighting a pixel value of the overlapping pixel by using a weighting factor specified from the weighting factor array, and a step of acquiring a combined image by combining the picked-up images. A pixel value of each overlapping pixel in the combined image is obtained by using weighted pixel values in two or more picked-up images.