Image Processing Method Using Block Segmentation for Resolution and Speed
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing methods using machine learning models struggle to produce high-resolution output images due to the down-sampling of input images, which results in lower resolution and information loss.
Innovation Solution
The method involves generating multiple second grayscale images by dividing a first grayscale image, reducing the number of pixels, and then upsampling these images using a machine learning model to produce high-resolution output images while minimizing information loss and reducing computational load.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If a down-sampled luminance image is used as input to reduce calculation amount, then processing speed is improved, but output image resolution deteriorates
Solution Approach 1:
The input image is divided into multiple blocks (e.g., 2x2 or 4x4 blocks), and each block is processed independently through the machine learning model. This segmentation allows the model to work with smaller down-sampled blocks while the final output is reconstructed by combining processed blocks, achieving both speed improvement and resolution maintenance.
Solution Approach 2:
The patent introduces a spatial dimension solution by dividing the image into multiple blocks and processing them separately. Instead of processing one large down-sampled image, the system processes multiple smaller blocks and reconstructs the full-resolution output by arranging processed blocks in their original spatial positions, effectively adding a spatial reconstruction dimension.
2Loss of energy
If a down-sampled image is used as input, then computational load is reduced, but information loss increases
Solution Approach 1:
By segmenting the image into blocks and processing each block separately with the machine learning model, the system maintains more information in each processed unit while reducing overall computational load through efficient block-wise processing. The segmentation preserves local image characteristics that would be lost in full-image down-sampling.
Solution Approach 2:
The patent applies partial action by processing only necessary blocks or regions of the image through the computationally intensive machine learning model, while other regions may use simpler processing. This selective approach reduces overall computational load while maintaining information quality in critical areas.
Data Source
AI summary
An image processing method includes generating, by dividing a first grayscale image, a plurality of second grayscale images where each has less number of pixels than that of the first grayscale image, and generating a plurality of third grayscale images upscaled by inputting the plurality of second grayscale images to a machine learning model.