Image Processing Apparatus for Endoscope Depth of Field
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Endoscope imaging systems face challenges in achieving a deep-focus image due to the small pixel pitch and high number of pixels, leading to a decreased depth of field and deteriorated image quality, especially when increasing the F-number to maintain depth of field, which results in noise increase and reduced imaging performance.
Innovation Solution
An image processing apparatus that acquires multiple images in different in-focus states, sets reference points, estimates distance information based on pixel values, and generates additional information such as lesion size or in-focus direction to enhance image processing and depth of field, utilizing techniques like axial chromatic aberration or time division to acquire images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the number of pixels of the imaging element is increased, then the resolution is improved, but the depth of field decreases
Solution Approach 1:
The patent divides the imaging process into multiple captures at different focal planes. Instead of relying on a single image from a high-resolution sensor with narrow depth of field, the system captures multiple images at different focus distances and then synthesizes them computationally. This segmentation of the imaging process allows the system to achieve both high resolution (through the high pixel count sensor) and extended depth of field (through multi-plane capture and synthesis).
Solution Approach 2:
The patent introduces the focal plane dimension as a new degree of freedom. By capturing images at multiple different focal planes (adding the depth dimension to the traditional 2D image plane), the system creates a multi-dimensional image dataset that can be processed to achieve both high resolution and extended depth of field simultaneously. This dimensional approach allows computational synthesis to reconstruct sharp images across the entire depth range.
2Length of stationary object
If the F-number of the optical system is increased, then the depth of field is improved, but the image quality deteriorates due to noise increase and optical system darkening
Solution Approach 1:
The patent replaces the mechanical/optical solution (using a high F-number aperture to increase depth of field) with a computational approach. Instead of relying on optical systems to achieve extended depth of field through aperture control, the system uses computational image synthesis to achieve the same effect. This substitution allows the optical system to operate at lower F-numbers (maintaining good image quality) while still achieving extended depth of field through software processing of multiple images at different focal planes.
3Length of stationary object
If the F-number of the optical system is increased, then the depth of field is improved, but the imaging performance deteriorates due to diffraction effect
Solution Approach 1:
The patent replaces the optical mechanism (high F-number aperture) with a computational method to achieve extended depth of field. By capturing multiple images at different focal planes and synthesizing them computationally, the system eliminates the need for high F-number aperture settings that cause diffraction. This allows the optical system to operate at optimal aperture values for minimizing diffraction effects while still achieving extended depth of field through image processing.
Data Source
AI summary
An image processing apparatus includes an image acquisition section that acquires a plurality of images that differ in in-focus state, a reference point setting section that performs a reference point setting process on each of the plurality of images, the reference point setting process setting a reference point that is set to an attention area, a distance estimation section that estimates distance information about a distance to a corresponding point based on a pixel value corresponding to the reference point, the corresponding point being a point in real space that corresponds to the reference point, and an additional information generation section that generates additional information based on the estimated distance information, the additional information being information that is added to the attention area to which the reference point is set.