Dynamic Parameter Control for Image Detection in Varying Scenes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image-taking systems, such as digital cameras, face challenges in maintaining effective detection performance under varying image-taking conditions, as the technique of adjusting threshold values based on brightness is insufficient in adapting to changes in image conditions.
Innovation Solution
An image-taking system that incorporates an objective body detecting means, a scene determining means, and a parameter control means to adjust parameters such as gain, threshold values, noise reduction, contour correction, and γ curve based on scene determination and timing, enhancing detection performance by integrating multiple detection results and adapting to changing conditions.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the threshold value is changed according to the brightness of the objective body, then the face detecting performance is improved, but the detecting performance cannot sufficiently deal with the change in the image-taking conditions
Solution Approach 1:
The patent applies parameter changes by dynamically adjusting multiple detection parameters (threshold value, gain, degree of noise reduction, degree of contour correction, γ curve) based on scene determination. Instead of only changing the threshold value according to brightness, the system changes multiple parameters according to determined scenes (backlight, night, underwater, etc.), thereby improving both detection accuracy and adaptability to various image-taking conditions.
Solution Approach 2:
The patent implements dynamics by making the detection parameters adjustable and changeable according to real-time scene determination. The parameter control means dynamically selects and adjusts parameters based on the currently determined scene, allowing the detection system to adapt flexibly to changing image-taking conditions rather than using fixed parameters.
2Measurement precision
If multiple parameters are used for detection, then the detecting performance is improved, but the complexity of the system increases
Solution Approach 1:
The patent uses parameter changes to manage complexity by adjusting detection parameters (threshold value, gain, noise reduction degree, contour correction degree, γ curve) based on scene determination. This approach improves detection accuracy across different conditions without requiring fundamentally different detection systems for each scene type.
Solution Approach 2:
The patent applies universality by creating a single detection system that can handle multiple scenes through parameter adjustment. The same objective body detection means uses different parameter sets for different scenes (backlight, night, underwater, etc.), making one system perform multiple detection functions rather than requiring separate systems for each condition.
3Adaptability or versatility
If the parameter is adjusted according to scene determination, then the detection performance under changing conditions is improved, but the processing time increases
Solution Approach 1:
The patent applies preliminary action by pre-defining multiple parameter sets corresponding to different scenes (backlight, night, underwater, etc.). The scene determining means quickly identifies the current scene type, and the parameter control means immediately selects the corresponding pre-prepared parameter set, avoiding time-consuming real-time parameter optimization calculations.
Solution Approach 2:
The patent uses feedback by implementing a closed-loop system where the scene determining means continuously monitors image characteristics, determines the current scene, and the parameter control means adjusts parameters based on this determination. This feedback mechanism allows the system to adapt to changing conditions efficiently by using scene information to guide parameter selection.
Data Source
AI summary
When detecting an objective body from a taken image, a scene of the image is determined, and a parameter to be used in detecting the objective body is controlled according to result of the determination.