Image Rotation and Bounding Box Cropping for Atlas Space Optimization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing methods, such as those used in Photoshop, result in large fragmented images that slow down loading speed and increase memory consumption, especially for complex images, and require numerous cropping steps, leading to long response times and inefficient use of atlas space.
Innovation Solution
A method that rotates images to different angles, calculates the bounding box area for each angle, and removes background pixels outside the smallest bounding box, reducing image size and improving loading speed and memory efficiency by generating smaller atlases with improved space utilization.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If images are divided into different image units using existing methods, then control for single image units is achieved, but the fragmented images become large in size which slows down loading speed and increases memory consumption
Solution Approach 1:
The patent extracts only the necessary image content by calculating the minimum bounding box that encloses all non-background pixels. This removes unnecessary background pixels and surrounding space, extracting only the essential image data while maintaining the ability to control individual image units.
Solution Approach 2:
The patent changes the parameter of image representation by storing only the bounding box coordinates and the cropped image content, rather than storing the entire original image canvas. This parameter transformation significantly reduces the quantity of data while preserving the functional capability of image unit control.
2Ease of manufacture
If existing image processing methods are used, then images can be processed, but the response time is long due to numerous cropping steps
Solution Approach 1:
The patent performs preliminary action by pre-calculating the minimum bounding box coordinates before actual image cropping. This preliminary calculation identifies the optimal crop region in advance, allowing subsequent cropping operations to be executed efficiently without requiring multiple trial cropping steps.
Solution Approach 2:
The patent replaces the mechanical iterative cropping process with a direct computational approach. Instead of repeatedly cropping and checking image units, the system uses pixel analysis to directly compute the minimum bounding box coordinates, substituting a mathematical calculation for a mechanical trial-and-error process.
3Device complexity
If images are processed without rotation optimization, then processing is simpler, but atlas space utilization is inefficient
Solution Approach 1:
The patent introduces dynamics by allowing images to be rotated to different angles before calculating the minimum bounding box. This dynamic rotation capability enables the system to find the optimal orientation that minimizes the bounding box area, thereby improving atlas space utilization while maintaining processing feasibility through automated angle optimization.
Solution Approach 2:
The patent changes the orientation parameter of images by rotating them to optimal angles before bounding box calculation. This parameter transformation allows the same image content to be represented with a smaller bounding box area when viewed from the optimal angle, improving space utilization in the atlas without requiring complex processing of multiple orientations simultaneously.
Data Source
AI summary
A method is provided. The method includes: obtaining a picture to be processed, where the picture to be processed includes a plurality of pixels, and the plurality of pixels comprise first pixels for forming an image and second pixels for forming an image background; rotating the picture to be processed, where for each rotation angle, an intermediate picture is obtained; selecting at least two pictures from the picture to be processed and several intermediate pictures for calculating an area of a bounding box surrounding the image respectively; and removing second pixels outside the bounding box in a picture with the smallest area of bounding box to obtain a processed picture.