Image Processing Device Background Texture Suppression
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing subject detection techniques in image processing devices face challenges when the background has a complex texture pattern, leading to inaccurate detection and potential generation of unnatural edges, especially when using convolutional neural networks.
Innovation Solution
The proposed image processing device adjusts frequency components or pixel values of image areas to reduce the influence of background textures, employing low-pass filtering, aperture control, and focus adjustments to enhance subject detection accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If subject detection is performed using conventional methods on images with complex background textures, then detection speed is maintained, but detection accuracy deteriorates due to confusion between subject and background textures
Solution Approach 1:
The patent extracts and removes frequency components corresponding to background textures from the image before subject detection. By separating the frequency domain representation of the image and eliminating components associated with background patterns, the system reduces the harmful influence of background textures on subject detection accuracy while maintaining detection speed.
Solution Approach 2:
The patent introduces frequency component adjustment as an intermediary processing step between image acquisition and subject detection. This intermediary process modifies the image data by adjusting frequency components to reduce background texture influence, thereby improving detection accuracy without directly affecting the detection algorithm's speed.
2Measurement precision
If subject area is extracted by cutting out from the original image, then subject detection focus is improved, but unnatural edges are generated reducing detection accuracy
Solution Approach 1:
The patent extracts frequency components corresponding to background textures and removes them from the image data. This frequency-domain extraction approach allows the system to focus on the subject area without generating unnatural edges, as the removal is performed in the frequency domain rather than through direct spatial cutting.
Solution Approach 2:
The patent changes the parameter space from spatial domain to frequency domain for the texture removal process. By operating in the frequency domain and adjusting frequency components rather than performing spatial cutting, the system maintains smooth transitions and avoids generating unnatural edges while still achieving the goal of removing background texture influence.
3Measurement precision
If frequency components or pixel values are adjusted to reduce background texture influence, then subject detection accuracy is improved, but image processing time increases
Solution Approach 1:
The patent applies partial action by selectively adjusting only the frequency components corresponding to background textures rather than processing the entire image uniformly. This selective approach reduces the computational burden while still achieving the goal of reducing background texture influence on subject detection accuracy.
Solution Approach 2:
The patent extracts and processes only the relevant frequency components associated with background textures, rather than performing comprehensive image processing. This extraction approach allows the system to reduce background texture influence with minimal additional processing time by focusing computational resources only on the necessary frequency components.
Data Source
AI summary
An imaging device acquires an input image using a lens unit and an imaging element and detects a subject. The imaging device calculates a reliability of detection of a subject and compares the reliability with a threshold value. When the reliability of detection of a subject is less than the threshold value, the imaging device performs a defocus calculating process and a background area determining process. The imaging device performs a low-pass filtering process on the determined background area, decreases a high-frequency component in the background area, and then detects a subject again.