Multi-pass Image Compression via Object Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Standard compression techniques, especially lossless compression, often result in larger file sizes and suffer from artifacts when decompressed, failing to efficiently compress data while preserving quality.
Innovation Solution
A multi-pass compression method using machine learning to identify and optimize the compression of specific objects within data, such as images or videos, by applying unique compression treatments to each object and background, allowing for smaller file sizes without artifacts upon decompression.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If lossless compression is used, then data quality is preserved, but file size is larger
Solution Approach 1:
The patent segments the image into multiple regions of interest (ROIs) and applies different compression treatments to each region. Important regions use lossless compression to preserve data quality, while less important regions use lossy compression to reduce file size, thereby resolving the contradiction between quality preservation and file size reduction.
Solution Approach 2:
The patent applies local quality by using different compression algorithms and quality levels for different regions of the image. Critical areas maintain high fidelity with lossless compression, while non-critical areas accept higher compression ratios, achieving both quality preservation where needed and file size reduction overall.
2Quantity of substance
If lossy compression is used, then file size is smaller, but compression artifacts appear
Solution Approach 1:
By segmenting the image and applying lossy compression only to non-critical regions while using lossless compression for important regions, the patent minimizes the visibility of compression artifacts while achieving overall file size reduction.
Solution Approach 2:
The patent converts the potential harm of compression artifacts into a benefit by strategically applying lossy compression only where artifacts are least noticeable, while protecting critical regions from artifact degradation through lossless compression.
3Productivity
If standard compression treatment is applied to all data, then processing is simple, but compression efficiency is poor
Solution Approach 1:
The patent divides the compression process into multiple stages: ROI identification, region segmentation, and differential compression application. This segmentation enables better compression efficiency by treating different regions differently, while the automated nature of the process manages processing complexity.
Solution Approach 2:
The patent performs preliminary actions by identifying and marking regions of interest before the actual compression process. This preliminary segmentation and classification enables more efficient compression in subsequent steps while automating the complexity management.
Data Source
AI summary
Introduced here is a technique to create small compressed image files while preserving data quality upon decompression. Upon receiving an uncompressed data, such as an image, a video, an audio, and/or a structured data, a machine learning model identifies an object in the uncompressed data such as a house, a dog, a text, a distinct audio signal, a unique data pattern, etc. The identified object is compressed using a compression treatment optimized for the identified object. The identified object, either before or after the compression, is removed from the uncompressed data. The uncompressed data with the identified object removed is compressed using a standard compression treatment.