Image Cropping with Target and Exclusion Frames
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image cutting methods often result in incomplete key information or excessive interference information, leading to suboptimal cutting effects.
Innovation Solution
An image cutting method that identifies and separates objects into target frames (key information) and exclusion frames (interference information) using object recognition models, allowing for precise cutting based on determined target areas.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If center cutting method is used to cut edge areas and reserve central area, then cutting process is simple, but key information may be incomplete or interference information is retained
Solution Approach 1:
The patent segments the image into multiple candidate areas based on different cutting strategies (center cutting, edge cutting, corner cutting). Each candidate area is evaluated separately, and the best one is selected. This segmentation approach allows the system to maintain simple cutting processes while improving cutting quality by choosing from multiple pre-defined cutting patterns.
Solution Approach 2:
The patent performs preliminary classification of objects in the image before cutting. It identifies key objects (such as faces, text, important elements) and pre-determines their positions. Based on this preliminary analysis, the system selects the most appropriate cutting strategy that preserves key information while removing interference, thus improving cutting effect without complicating the actual cutting process.
2Productivity
If traditional image cutting methods are used, then processing speed is fast, but key information may be lost or interference information remains
Solution Approach 1:
The patent performs preliminary object recognition and classification before the actual cutting operation. It identifies key objects (faces, text, important elements) and determines their positions in advance. This preliminary analysis enables the system to quickly select the optimal cutting strategy without sacrificing cutting speed, as the heavy lifting of information identification is done before the actual cutting decision.
Solution Approach 2:
The patent uses feedback from object recognition results to guide the cutting process. It analyzes the distribution and importance of objects in the image, then adjusts the cutting strategy accordingly. This feedback mechanism ensures that key information is preserved while maintaining efficient processing, as the system adapts its cutting approach based on the actual content of the image rather than using a fixed method.
Data Source
AI summary
The present disclosure provides an image cutting method performed by a computer device. The method includes: determining an object frame in which an object in a first image belonging to a first type is located as a first target frame and determining an object frame in which an object in the first image belonging to a second type is located as an exclusion frame; determining a target area in the first image, the target area comprising the first target frame and not comprising the exclusion frame; and cutting the first image based on the target area to obtain a second image comprising the first target frame and not comprising the exclusion frame. Using the above method, apparatus, computer device, and storage medium, the image cutting effect and the image cutting speed can be improved.