Image Segmentation Using Position-Aware Weight Distribution
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current image segmentation techniques fail to effectively segment garment images due to complex backgrounds and layouts, as they are not applicable for images with unclear fashion models or intricate textures.
Innovation Solution
An image segmentation method that performs classification based on the position of the subject in the image, selecting a subject position template with a weight distribution field to differentiate between foreground and background pixels, allowing for improved segmentation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If segmentation techniques based on significant region detection are used, then segmentation effect is good when image has clear background and simple layout, but segmentation fails when image has complex background or complex layout
Solution Approach 1:
The patent changes the parameter of background complexity handling by introducing a weight distribution field that dynamically adjusts pixel weights based on their likelihood of belonging to foreground or background. This allows the segmentation algorithm to adapt to varying background complexities without changing the fundamental segmentation approach.
Solution Approach 2:
The patent introduces a weight distribution field as an intermediary between the image input and segmentation output. This field acts as a mediator that softens the binary classification decision by providing probabilistic weights, enabling more nuanced handling of complex backgrounds and layouts.
2Measurement precision
If segmentation techniques based on face detection are used, then segmentation is suitable for images with clear face and simple posture, but segmentation fails when image has no fashion model or complex posture
Solution Approach 1:
The patent changes the parameter of subject position adaptability by using multiple pre-defined position parameters (e.g., center, left, right, top, bottom) with associated weight distribution fields. This allows the system to adapt to various subject positions and postures by selecting and applying the appropriate position template for each image.
3Measurement precision
If segmentation techniques based on image connectivity are used, then segmentation is suitable for images with clear background, simple layout and garment with little texture, but segmentation fails when image has complex background or complex layout
Solution Approach 1:
The patent changes the parameter of layout complexity handling by introducing position-aware weight distribution fields that consider both spatial location and background complexity. This allows the segmentation to adapt to complex layouts by dynamically adjusting the importance of different image regions based on their position and local characteristics.
Data Source
AI summary
A system and method involving performing image classification on the image according to a position of a subject in the image; selecting, from a plurality of subject position templates, a subject position template for the image according to a result of the image classification, wherein each of the plurality of subject position templates is associated with a pre-defined position parameter, and each of the plurality of subject position templates is configured with a weight distribution field according to the pre-defined position parameter, the weight distribution field representing a probability that each pixel in the image belongs to a foreground or a background; and performing image segmentation according to the weight distribution field in the selected subject position template to segment the subject from the image.