Image Processing Apparatus Coordinate Transformation Precision
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing apparatuses face challenges in achieving high-precision coordinate transformation for image deformation due to limitations in hardware circuits and software performance, leading to insufficient calculation precision, especially for high-resolution images or large deformations.
Innovation Solution
An image processing apparatus that includes a divisor calculation unit to determine the minimum and maximum values of the divisor, a coefficient calculation unit to set the divisor range within an allowable error, and a parameter adjustment unit to calculate a normalization parameter, which is used to adjust the deformation parameter for improved precision in coordinate transformation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high-precision coordinate calculation is performed using hardware circuits, then calculation precision is improved, but the amount of circuits and operating frequency requirements increase
Solution Approach 1:
The patent changes the parameter representation by introducing a normalization parameter that transforms the division operation parameters. By calculating the minimum value of the divisor and using it to normalize the division operation, the system achieves high-precision coordinate calculation without requiring complex hardware circuits. The normalization parameter adjusts the dividend and divisor relationship to maintain precision while simplifying the computational requirements.
2Measurement precision
If high-precision coordinate calculation is performed using software, then calculation precision is improved, but processing speed decreases
Solution Approach 1:
The patent performs preliminary calculation of the normalization parameter based on the minimum value of the divisor before the actual coordinate transformation. This pre-computed normalization parameter is then reused in subsequent division operations, eliminating the need for repeated complex calculations. The preliminary action of computing the normalization parameter once enables faster processing while maintaining high precision in coordinate calculations.
3Productivity
If approximation calculation is performed in the neighborhood of a pixel, then calculation amount is reduced, but calculation precision deteriorates
Solution Approach 1:
The patent transforms the approximation approach by changing the parameters of the division operation through normalization. Instead of approximating the result of division, the system normalizes the divisor to a standard range (with minimum value of 1), which allows for more accurate calculation while reducing the computational complexity. This parameter transformation enables precise coordinate calculation without relying on neighborhood approximation methods.
4Measurement precision
If divisor and dividend are doubled until MSB of divisor becomes 1 for each division, then division accuracy is improved, but calculation amount and power consumption increase
Solution Approach 1:
The patent performs the divisor normalization operation in advance by calculating the minimum value of the divisor and determining the normalization parameter before actual image processing. This pre-computed normalization parameter ensures that the divisor will have an MSB of 1 during division operations, guaranteeing division accuracy without requiring repeated doubling operations during processing. The preliminary normalization eliminates the need for iterative adjustments during each division operation.
Data Source
AI summary
An image processing apparatus that generates an output image by deforming an input image by coordinate transformation, comprises a divisor calculation unit calculates a minimum value and a maximum value of a divisor, used in division operation of coordinate transformation; a coefficient calculation unit obtains a range of the divisor in a way that a result of the division operation becomes equal to or smaller than an allowable error, and calculates a normalization parameter by dividing a minimum value of the range by the minimum value of the divisor; a parameter adjustment unit calculates an adjustment parameter used in the coordinate transformation, and outputs the adjustment parameter; and an image deformation processing unit performs transformation processing of the coordinate information using the adjustment parameter, and outputs a result of the transformation processing.