Iterative Subject Detection Region Adjustment
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional subject detection technologies face challenges with inaccurate image detection, particularly in capturing moving objects within images.
Innovation Solution
A subject detection method and apparatus that involve detecting a moving object region on a captured image, sliding a box to identify candidate regions, and iteratively adjusting the size of the first region based on its proportion to the moving object region, until a threshold is reached, to determine a target region for improved accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional subject detection technology is used, then the detection process is simple, but the detection accuracy is low
Solution Approach 1:
The patent applies dynamics by implementing an iterative adjustment mechanism where the detection region is dynamically refined through multiple cycles. The sliding box iteratively adjusts its position and size based on proportion calculations, transforming a static detection process into a dynamic one that progressively improves accuracy until convergence criteria are met.
Solution Approach 2:
The patent implements feedback by calculating the proportion of the sliding box region relative to the moving object region and using this feedback to guide subsequent adjustments. The system continuously monitors the detection quality through proportion calculations and adjusts the detection region accordingly, creating a closed-loop feedback system that enhances detection accuracy.
2Measurement precision
If the detection region size is fixed, then the processing speed is fast, but the detection precision is insufficient
Solution Approach 1:
The patent applies partial action by performing iterative adjustments only when necessary to improve precision. The system calculates proportions and adjusts the sliding box region selectively based on whether the current region adequately captures the moving object, avoiding unnecessary iterations and maintaining processing efficiency while improving detection precision.
Solution Approach 2:
The patent implements parameter changes by dynamically adjusting the size and position parameters of the detection region based on calculated proportions. The system modifies the sliding box parameters iteratively, changing them from fixed to variable values that adapt to the actual moving object characteristics, thereby improving detection precision without excessive computational overhead.
3Measurement precision
If a single candidate region is selected, then the processing is efficient, but the detection accuracy is limited
Solution Approach 1:
The patent applies preliminary action by first identifying a moving object region through initial detection, then using this preliminary result to guide the subsequent sliding box placement. The system performs preliminary proportion calculations and uses these results to inform the iterative adjustment process, ensuring that each subsequent action is based on accurate prior information.
Solution Approach 2:
The patent implements segmentation by dividing the detection process into distinct stages: initial moving object detection, sliding box placement, proportion calculation, and iterative adjustment. This segmentation allows the system to handle complex detection tasks through a series of simpler, manageable steps, improving accuracy without overwhelming computational complexity.
Data Source
Figure 1
Figure 2
Figure 3~4
AI summary
A subject detection method includes as follows. A moving object region is obtained by detecting a moving object on a captured first image. A sliding box is obtained, a plurality of candidate regions are obtained by sliding the sliding box on the first image, and a first region is determined. The first region is one of the plurality of candidate regions which includes a part of the moving object region with a largest area among the plurality of candidate regions. A proportion of the first region is obtained. A second region is obtained by adjusting a size of the first region based on the proportion of the first region. A proportion of the second region is obtained, the first region is replaced with the second region, the proportion of the first region is replaced with the proportion of the second region, and it is returned to performing the operation of adjusting the size of the first region based on the proportion of the first region until a number of times of the iterative adjustments reaches a threshold, and a region obtained by the last iterative adjustment is determined as a target region.