Capsule Endoscope Depth Map Image Sharpening
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Capsule endoscopes face challenges in capturing sharp images of both close and distant objects within the gastrointestinal tract due to limited camera depth of field, which is exacerbated by fixed focus and size/power constraints, making it difficult to achieve high-quality imaging across varying object distances.
Innovation Solution
The method involves capturing structured-light images and regular images, deriving a depth map to determine a deconvolution kernel based on camera parameters, and applying filters to improve image sharpness, allowing for optimized image processing and sharpening of target regions within the gastrointestinal tract.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If fixed focus camera is used in capsule endoscope, then device complexity is reduced, but image quality for objects at different distances deteriorates
Solution Approach 1:
The system performs preliminary actions by capturing structured light images to derive depth information before processing regular images. The depth map is calculated in advance to determine appropriate deconvolution kernels for different regions, enabling post-processing sharpening without requiring complex real-time focusing mechanisms.
Solution Approach 2:
The patent replaces mechanical focusing systems with computational image processing. Instead of using variable focus lenses or multiple lenses to achieve sharp images at different depths, the system uses fixed focus cameras combined with depth-based deconvolution algorithms to achieve selective sharpness in post-processing.
2Manufacturing precision
If variable focus implementation is added to capsule endoscope, then image quality for different distances is improved, but device complexity and power consumption increase
Solution Approach 1:
The patent substitutes mechanical variable focus systems with computational methods. Fixed focus cameras capture images at all distances, and depth-based deconvolution algorithms selectively sharpen regions at different depths, eliminating the need for moving parts or complex lens mechanisms.
Solution Approach 2:
The system introduces depth maps derived from structured light images as an intermediary to guide the image processing. This depth information acts as a mediator between the fixed focus camera and the deconvolution algorithm, enabling selective sharpening without requiring the camera to physically adjust focus.
3Manufacturing precision
If depth-based filtering is applied to improve sharpness, then image quality is enhanced, but processing time and computational resources increase
Solution Approach 1:
The system segments the image into multiple regions based on depth information from the depth map. Each region is processed with appropriate deconvolution kernels tailored to its distance, allowing parallel processing and optimization of computational resources rather than processing the entire image uniformly.
Solution Approach 2:
The patent applies local quality enhancement by using different deconvolution kernels for different regions of the image based on their depth. Instead of applying a uniform processing approach, the system tailors the sharpening strength and kernel parameters to local depth characteristics, improving efficiency and effectiveness.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This approach enables the production of high-quality images for both near and far objects in a single frame, enhancing visualization of gastrointestinal tract mucosa and reducing the need for active navigation, while conserving system resources like battery energy and memory.
Implementation Method 1
One or more structured-light images are received, where said one or more structured-light images are captured using the imaging apparatus by projecting the body lumen with structured light
Implementation Method 2
A first processed target region is generated by applying the filter to the target region to improve sharpness of the target region. The filter may correspond to a deconvolution kernel, and parameters of the filter are designed based on the target distance and the camera parameters
Data Source
AI summary
A method and apparatus for processing gastrointestinal (GI) images are disclosed. According to this method, a regular image is received, where the regular image is captured using an imaging apparatus by projecting non-structured light onto a body lumen when the imaging apparatus is in the body lumen. One or more structured-light images captured using the imaging apparatus by projecting the body lumen with structured light are received. A target distance for a target region in the regular image is derived based on said one or more structured-light images. A filter is determined based on the target distance and camera parameters associated with the imaging apparatus. A first processed target region is generated by applying the filter to the target region to improve sharpness of the target region. A first processed regular image comprising the first processed target region is provided.