Person State Detection Range Setting for Spherical Images
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing techniques face challenges in achieving both reduced detection time and increased accuracy for detecting the state of a person from a screen, particularly in spherical images, due to issues such as image distortion, reduced pixel count, and division of the person across image boundaries.
Innovation Solution
The method involves identifying the position of a person's region within the image and setting an object detection range based on this position, duplicating image portions to maintain pixel count, and performing detection only on this range, thereby reducing detection time without compromising accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If deep learning is used to detect person state from entire screen, then detection accuracy is improved, but detection time increases
Solution Approach 1:
The patent divides the screen into multiple regions and performs detection in a focused region rather than the entire screen. The system identifies a person's location first, then sets a specific detection range around that person, segmenting the detection task from the full screen to reduce processing time while maintaining accuracy on the target subject.
Solution Approach 2:
The patent applies different processing qualities to different regions. High-resolution detection is applied only to the region containing the person, while other regions receive less processing. This local quality approach ensures accurate detection of the person's state without unnecessarily processing the entire screen at full resolution.
2Area of stationary object
If spherical image is used to capture wide angle, then coverage area is improved, but image distortion and pixel count reduction occur
Solution Approach 1:
The patent segments the spherical image into multiple planar images through development. This transformation divides the distorted spherical coordinate system into manageable planar regions, reducing distortion effects in each segment while preserving the wide-angle coverage capability of the original spherical capture.
Solution Approach 2:
The patent transforms the three-dimensional spherical image into a two-dimensional developed image. This dimensional change allows the wide-angle field of view to be represented in a planar format that is more suitable for standard image processing operations, reducing the negative effects of spherical distortion.
3Reliability
If detection is performed on entire screen, then comprehensive detection is improved, but detection speed decreases
Solution Approach 1:
The patent segments the detection process into two stages: first detecting the person's location in the entire screen, then performing detailed state detection only in the region surrounding the person. This segmentation maintains comprehensive detection capability while significantly improving detection speed by limiting detailed analysis to a smaller region.
Solution Approach 2:
The patent performs preliminary detection of the person's location before conducting detailed state detection. This preliminary action identifies the region of interest, allowing subsequent detection to focus only on that area, thereby improving overall detection speed without compromising the comprehensiveness of the final result.
Data Source
AI summary
An image processing method includes acquiring an image; identifying a position of a region of a person in the acquired image; setting an object detection range based on the identified position of the region of the person; and detecting a state of the person based on the set object detection range.