Image Segmentation Blending via Reference Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image segmentation technologies struggle to perform consistently across various devices due to differences in hardware and camera settings, leading to inconsistent object classification and 'flicker' in video playback.
Innovation Solution
A method that applies a blending process to input image segmentation data and reference image segmentation data, using input image feature data and reference image feature data to generate blended image segmentation data, which characterizes a final segmentation of the input image.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If image segmentation is performed using standard machine learning processes, then object classification can be achieved, but segmentation consistency deteriorates across different devices due to hardware and camera setting variations
Solution Approach 1:
The patent introduces reference images as an intermediary element that mediates between the input image and segmentation results. By comparing the input image against multiple reference images with known segmentations, the system creates a bridge that compensates for device-specific variations. The reference images serve as a common reference frame that enables consistent segmentation across different devices, effectively resolving the contradiction between achieving accurate segmentation and maintaining consistency across varying hardware conditions
Solution Approach 2:
The patent changes the parameters used for segmentation by incorporating multiple reference images with different characteristics into the segmentation process. Instead of relying on a single segmentation model trained on generic data, the system uses parameter variations from multiple reference images (different lighting conditions, angles, resolutions) to adapt the segmentation to specific device characteristics. This parameter change approach allows the system to maintain high segmentation accuracy while improving consistency across different devices
2Reliability
If reference images from multiple sources are integrated, then segmentation consistency improves, but processing complexity increases
Solution Approach 1:
The patent applies preliminary action by pre-processing and organizing reference images before they are used for segmentation. Reference images are collected, labeled, and structured in advance according to specific criteria (device types, lighting conditions, etc.). This preliminary organization creates a ready-to-use reference library that can be quickly queried and applied during actual segmentation tasks, reducing the processing complexity at runtime while maintaining the consistency benefits of using multiple reference images
Solution Approach 2:
The patent segments the reference image set into multiple categories or groups based on specific characteristics (e.g., device type, lighting condition, resolution). This segmentation of reference images allows the system to selectively apply only the relevant reference images for a given input image, rather than processing all reference images. By dividing the reference set into manageable segments, the system reduces processing complexity while maintaining segmentation consistency through targeted comparison
Data Source
AI summary
Methods, systems, and apparatuses for image segmentation are provided. For example, a computing device may obtain an image, and may apply a process to the image to generate input image feature data and input image segmentation data. Further, the computing device may obtain reference image feature data and reference image classification data for a plurality of reference images. The computing device may generate reference image segmentation data based on the reference image feature data, the reference image classification data, and the input image feature data. The computing device may further blend the input image segmentation data and the reference image segmentation data to generate blended image segmentation data. The computing device may store the blended image segmentation data within a data repository. In some examples, the computing device provides the blended image segmentation data for display.