Image Search Subject Isolation via Object Masking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image search platforms struggle to focus on the object of interest when multiple objects are present in the input image, leading to poor search results due to noise from other objects, and zooming techniques often reduce image quality and fail to isolate the subject object effectively.
Innovation Solution
A system processes image data to detect objects, generates masks for segmentation, determines a subject object by intersecting masks with a focus region, and excludes other objects from the search input, allowing for precise targeting of the subject object in image searches.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If all objects in the input image are used for search, then the search covers more objects, but the search results are contaminated by noise from unrelated objects
Solution Approach 1:
The patent segments the input image into multiple object regions by detecting individual objects and generating separate masks for each. This allows the system to process and search each object independently, enabling selective inclusion of relevant objects while excluding unrelated noise objects from the search query.
Solution Approach 2:
The patent extracts the focus region containing the subject object from the full input image by detecting object boundaries and selecting the region of interest. This extraction removes unrelated objects and background noise, creating a cleaned search input that focuses only on the relevant object for accurate search results.
2Measurement precision
If the user zooms in on the object of interest, then the search focuses more on the subject object, but the image quality deteriorates and overlapping objects cannot be isolated
Solution Approach 1:
Instead of relying on zooming, the patent segments the image into distinct object masks based on detected object boundaries. This segmentation allows precise isolation of the subject object even when objects are overlapping or at the same depth, maintaining image quality while achieving accurate search focus.
Solution Approach 2:
The patent introduces object detection and mask generation as intermediary steps between the input image and the search query. This intermediary processing layer identifies and isolates the subject object without requiring optical zooming, thereby preserving image quality while achieving precise search focus.
3Device complexity
If conventional zooming is used to focus on the subject object, then the search input is simplified, but objects at the same depth cannot be separated and search results remain contaminated
Solution Approach 1:
The patent segments objects based on their spatial boundaries and depth information rather than relying on 2D zooming. By detecting object boundaries and generating corresponding masks, the system can separate objects at the same depth that would otherwise be indistinguishable in a zoomed-in 2D view, improving search accuracy without excessive complexity.
Solution Approach 2:
The patent transitions from 2D image processing to 3D spatial understanding by detecting object depth information and boundaries. This dimensional extension allows the system to separate and isolate objects based on their spatial arrangement in three-dimensional space, enabling accurate search focus even when objects overlap in the 2D image plane.
Data Source
AI summary
In implementations of search input generation for an image search, a computing device can capture image data of an environment scene that includes multiple objects. The computing device implements a search input module that can detect the multiple objects in the image data, and initiate a display of a selectable indication for each of the multiple objects. The search input module can then determine a subject object from the detected multiple objects, and generate the subject object as the search input for the image search.