Mesh-Based Image Downsampling for Bloom Effect Efficiency

Solution Overview

Problem

Existing image processing methods, such as the Bloom effect, require multiple downsampling and upsampling processes, leading to high performance overhead and low image processing efficiency.

Innovation Solution

An image processing method that constructs a mesh with multiple mesh cells of different sizes based on image parameters, downsamples the target image using these mesh cells to obtain multiple downsampled images, and then performs image fusion to obtain the target image.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Manufacturing precision

If multiple downsampling and upsampling processes are performed to achieve Bloom effect, then image processing quality is improved, but performance overhead increases and processing efficiency decreases

Engineering Contradiction:

Improveimage processing qualityVSAvoidprocessing efficiency

Core Design Contradiction:

Manufacturing precisionVSProductivity

Solution Approach 1:

The patent combines multiple downsampling and upsampling operations into a single mesh-based processing step. Instead of performing sequential downsampling, blurring, and upsampling operations multiple times, the invention uses a mesh structure to accomplish all these functions in one pass, thereby maintaining image quality while dramatically improving processing efficiency and reducing performance overhead.

Inventive Principle:

Principle #5Merging (Combining)

2Manufacturing precision

If multiple downsampling and upsampling processes are performed to achieve Bloom effect, then image processing quality is improved, but performance overhead increases

Engineering Contradiction:

Improveimage processing qualityVSAvoidperformance overhead

Core Design Contradiction:

Manufacturing precisionVSUse of energy by moving object

Solution Approach 1:

The patent combines multiple downsampling and upsampling operations into a single mesh-based processing step. Instead of performing sequential downsampling, blurring, and upsampling operations multiple times, the invention uses a mesh structure to accomplish all these functions in one pass, thereby maintaining image quality while dramatically improving processing efficiency and reducing performance overhead.

Inventive Principle:

Principle #5Merging (Combining)

3Illumination intensity

If traditional Bloom effect processing is used, then light and dark contrast is enhanced, but the number of rendering processes increases to at least six

Engineering Contradiction:

Improvelight and dark contrastVSAvoidnumber of rendering processes

Core Design Contradiction:

Illumination intensityVSDevice complexity

Solution Approach 1:

The patent combines multiple downsampling and upsampling operations into a single mesh-based processing step. Instead of performing sequential downsampling, blurring, and upsampling operations multiple times, the invention uses a mesh structure to accomplish all these functions in one pass, thereby maintaining image quality while dramatically improving processing efficiency and reducing performance overhead.

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The patent divides the image into a mesh structure with multiple mesh cells of different sizes. This segmentation allows different regions of the image to be processed at different levels of detail, enabling the Bloom effect to be achieved with fewer rendering processes while maintaining contrast enhancement in critical areas.

Inventive Principle:

Principle #1Segmentation

Data Source

PatentUS20250037232A1Image processing method and apparatus, electronic device, storage medium, and program product

Publication Date: 2025.01.30 TENCENT TECHNOLOGY (SHENZHEN) CO LTD

AI summary

This application provides an image processing method and apparatus, an electronic device, a computer-readable storage medium, and a computer program product. The method includes: obtaining an image parameter corresponding to a target image, and constructing, based on the image parameter, a mesh for downsampling the target image, the mesh including N mesh cells, the N mesh cells including at least mesh cells of different sizes, and N being a positive integer greater than 1; using the mesh cells in the mesh separately to downsample the target image to obtain N downsampled images; and performing image fusion on the N downsampled images to obtain a target image.