Adaptive Image Noise Reduction for Endoscopic Imaging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image processing technologies face challenges in accurately determining whether an image is in a stationary or moving state, leading to inadequate noise reduction processes that either attenuate high-frequency components or result in residual images.
Innovation Solution
An image processing device that estimates the noise amount and uses an inter-frame difference value to determine the image state, adaptively selecting between time-direction and spatial-direction noise reduction processes to maintain high-frequency components while reducing noise effectively.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Object-affected harmful factors
If time-direction noise reduction process is used, then noise is reduced effectively, but high-frequency components are attenuated
Solution Approach 1:
The patent applies dynamics by making the noise reduction process adaptive rather than static. The system dynamically switches between time-direction and spatial-direction noise reduction processes based on real-time determination of image state (stationary or moving). This allows the noise reduction strength and direction to change according to the actual image conditions, preventing both over-reduction of high-frequency components in moving images and insufficient noise reduction in stationary images.
2Manufacturing precision
If spatial-direction noise reduction process is used, then high-frequency components are preserved, but noise reduction effectiveness is reduced
Solution Approach 1:
The patent applies local quality by applying different noise reduction strategies to different local states of the image. Instead of uniformly applying one noise reduction process to the entire image, the system determines the state (stationary or moving) for each image and applies the appropriate noise reduction process (spatial-direction for moving, time-direction for stationary) locally to each region or frame, optimizing both noise reduction and detail preservation for each specific condition.
3Device complexity
If fixed threshold value is used for determining image state, then determination process is simple, but determination accuracy is insufficient
Solution Approach 1:
The patent applies parameter changes by using multiple parameters (inter-frame difference value, noise amount, evaluation value) instead of a single fixed threshold. The system calculates the inter-frame difference value between consecutive images, compares it with the estimated noise amount, and determines the image state based on this dynamic evaluation. This multi-parameter approach significantly improves determination accuracy while maintaining reasonable computational complexity.
Data Source
Figure 1
Figure 2~3
Figure 4
AI summary
An image processing device includes an evaluation value calculation section (321), an estimated noise amount acquisition section (322), a determination section (324), and a noise reduction processing section (325). The evaluation value calculation section (321) calculates an evaluation value that is used to determine whether or not an inter-frame state of an object within a captured image is a stationary state. The estimated noise amount acquisition section (322) acquires an estimated noise amount of the captured image. The determination section (324) determines whether or not the inter-frame state of the object is the stationary state based on the evaluation value and the estimated noise amount. The noise reduction processing section (325) performs a first noise reduction process (time-direction noise reduction process) on the captured image when it has been determined that the inter-frame state of the object is the stationary state, and performs a second noise reduction process that includes at least a spatial-direction noise reduction process on the captured image when it has been determined that the inter-frame state of the object is not the stationary state.