Image Scaling Filter Reuse for Mobile Device Computational Efficiency

Solution Overview

Problem

Existing image scaling algorithms for mobile devices are computationally expensive and result in low image quality due to the need for complex filter calculations, leading to visible artifacts such as pixilation and aliasing, especially when scaling images with high resolutions.

Innovation Solution

The proposed method involves cyclic reuse of filters by determining a filter set comprising α filters, where α is the quotient of the target resolution and a common denominator of the original and target resolutions, allowing for reduced computational effort and complexity by reusing filters up to k times when scaling images.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If complex filter calculations are used for image scaling, then image quality is improved, but computational cost and processing time increase significantly

Engineering Contradiction:

Improveimage qualityVSAvoidcomputational complexity

Core Design Contradiction:

Manufacturing precisionVSDevice complexity

Solution Approach 1:

The patent segments the filter calculation process by pre-calculating and storing filter coefficients in a lookup table, then retrieving and reusing them during image scaling. This divides the computational workload into a one-time calculation phase and multiple reuse phases, significantly reducing per-pixel computational complexity while maintaining image quality.

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent performs preliminary action by pre-calculating filter coefficients before the actual image scaling operation. The filter coefficients are computed once and stored in a lookup table, then retrieved and reused for all pixels during scaling. This preliminary computation eliminates the need for repeated complex filter calculations during image processing.

Inventive Principle:

Principle #10Preliminary action

2Manufacturing precision

If more filter coefficients are calculated for each pixel, then image quality improves, but processing time and power consumption increase

Engineering Contradiction:

Improveimage qualityVSAvoidprocessing time

Core Design Contradiction:

Manufacturing precisionVSLoss of time

Solution Approach 1:

The patent performs preliminary action by pre-calculating filter coefficients before the actual image scaling operation. The filter coefficients are computed once and stored in a lookup table, then retrieved and reused for all pixels during scaling. This preliminary computation eliminates the need for repeated complex filter calculations during image processing.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent uses copying by retrieving pre-calculated filter coefficients from a lookup table and reusing them for multiple pixels. Instead of recalculating filters for each pixel, the system copies the same filter coefficients from memory, significantly reducing processing time and power consumption while maintaining image quality.

Inventive Principle:

Principle #26Copying

3Productivity

If simple scaling techniques like nearest neighbor are used, then processing speed increases, but image quality deteriorates with visible artifacts

Engineering Contradiction:

Improveprocessing speedVSAvoidimage quality

Core Design Contradiction:

ProductivityVSManufacturing precision

Solution Approach 1:

The patent performs preliminary action by pre-calculating filter coefficients before the actual image scaling operation. The filter coefficients are computed once and stored in a lookup table, then retrieved and reused for all pixels during scaling. This preliminary computation eliminates the need for repeated complex filter calculations during image processing.

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS7848597B2Data processing

Publication Date: 2010.12.07 TELEFONAKTIEBOLAGET LM ERICSSON (PUBL)

AI summary

In an image scaling method, an image (10) having an original row/column resolution NO is scaled into a scaled image (30) having a target row/column resolution NT. A filter set comprising α filters is provided, where α being equal to a quotient of the target resolution NT and a common denominator k of the original resolution NO and the target resolution NT. The common denominator k is a positive number larger than one. For at least one row/column (16/18) of image elements (12), a filtered image element (32) of the scaled image (30) is formed, for each filter in the filter set, by filtering at least one image element (12) of the at least one row/column (16/18) using the filter. This image element formation is repeated up to k times over the at least row/column (16/18) to form a scaled row/column (36/38) of image elements (32) of the scaled image (30).