Rim-wise Function Estimation for Image Processing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for evaluating continuous functions on discrete image data, such as the Laplacian of Gaussian filter, require significant processing load, especially on battery-powered devices like cellular phones, due to the need to compute weighted sums for numerous pixel positions in images.
Innovation Solution
A method that estimates a continuous function by pre-computing and storing discrete weightings for sets of pixels, allowing for efficient evaluation by summing attribute values and applying weights, and modifying sums for adjacent pixel positions to reduce redundant calculations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If a continuous function is evaluated using discrete data points with weighted sums, then measurement precision is improved, but processing load increases
Solution Approach 1:
The image is divided into multiple blocks, and the function is evaluated separately for each block rather than for all pixels individually. This segmentation reduces the computational complexity by processing smaller subsets of data in parallel or sequentially, thereby improving processing speed while maintaining measurement precision through systematic coverage of the entire image.
Solution Approach 2:
Weighting factors are pre-calculated and stored in a lookup table before the actual function evaluation. This preliminary action eliminates the need to compute weighting factors during the main processing stage, significantly reducing the processing load while preserving the accuracy of the weighted sum calculation.
2Measurement precision
If weighting factors are calculated for each pixel position, then measurement precision is improved, but use of energy increases
Solution Approach 1:
Weighting factors are pre-calculated and stored in a lookup table before the actual function evaluation. This preliminary action eliminates the need to compute weighting factors during the main processing stage, significantly reducing the processing load while preserving the accuracy of the weighted sum calculation.
Solution Approach 2:
Instead of recalculating weighting factors for each pixel position, the patent uses pre-computed weighting factors that are copied and applied to different pixel positions. This copying approach maintains measurement precision while dramatically reducing energy consumption by avoiding redundant calculations.
3Measurement precision
If function evaluation is performed for all pixel positions, then measurement precision is improved, but loss of time increases
Solution Approach 1:
The image is divided into multiple blocks, and the function is evaluated separately for each block rather than for all pixels individually. This segmentation reduces the computational complexity by processing smaller subsets of data in parallel or sequentially, thereby improving processing speed while maintaining measurement precision through systematic coverage of the entire image.
Solution Approach 2:
The patent evaluates the function for representative pixel positions within blocks rather than for every single pixel. This partial action approach maintains the essential measurement precision needed for image analysis while significantly reducing the total processing time by avoiding redundant evaluations at every pixel position.
Data Source
Figure 1~2
Figure 3~4
Figure 5
AI summary
A method for analysing an image composed of pixels by estimating a continuous function of a selected pixel attribute, the continuous function being centred on a selected pixel position, the method comprising: determining a series of nested sets of pixels, each set being centred on a common pixel; estimating a discrete weighting for each set of pixels in accordance with the continuous function; mapping the sets of pixels onto pixels of the image with the common pixel mapped on to the selected pixel position; for each set of pixels, forming a combined value in dependence on the values of the selected pixel attribute for the pixels of the set in the image; and estimating the function as a sum of the combined values for each set weighted by the respective discrete weighting.