Diamond Clarity Grading via Multi-Focal Optical Imaging
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for assessing diamond clarity are subjective and lack repeatability, as they rely on human visual inspection, which can lead to inconsistent results due to factors like fatigue and varying judgments among gemologists.
Innovation Solution
A system and process utilizing optical image acquisition and processing techniques, including High Dynamic Range imaging, flat-field correction, and a neural network for analyzing axial view images of diamonds under controlled lighting, to objectively determine clarity by combining images from different focal depths and angles.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If visual inspection under 10x magnification by trained gemologists is used, then clarity assessment can be performed, but repeatability and consistency of results deteriorate due to subjective human judgement and fatigue
Solution Approach 1:
The patent replaces the mechanical/visual inspection system with an optical imaging system. A camera captures images of the diamond under controlled lighting conditions, and these images are processed by a neural network to determine clarity characteristics. This substitution eliminates human subjectivity and fatigue, providing consistent and repeatable measurements.
Solution Approach 2:
The patent creates optical copies (images) of the diamond at different magnifications and focal depths. These images serve as digital representations that can be analyzed without direct human visual inspection. The neural network processes these copied images to extract clarity information, ensuring consistent interpretation across different assessments.
2Measurement precision
If multiple images at different focal depths are acquired and combined, then measurement precision of internal defects is improved, but device complexity and processing time increase
Solution Approach 1:
The patent segments the diamond into different focal depth planes and captures images at each plane. By dividing the three-dimensional inspection space into multiple two-dimensional image planes, the system can systematically examine internal defects at different depths. The neural network then integrates information from these segmented images to provide comprehensive clarity assessment.
Solution Approach 2:
The patent transitions from two-dimensional surface inspection to three-dimensional volumetric inspection by acquiring images at multiple focal depths. This adds the depth dimension to the inspection process, enabling detection of internal defects that would be invisible in single-plane images. The neural network processes this multi-dimensional data to extract clarity characteristics.
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 provides a consistent and reliable method for assessing diamond clarity, reducing human error and improving repeatability, while also reducing the time and cost associated with training gemologists and producing master stone sets.
Implementation Method 1
an optical image acquisition device for acquiring one or more plurality of axial view images of a diamond with different focus depths
Implementation Method 2
Different focus depths corrected with the refractive index of the diamond. The apparent focus depth D apparent for focusing may be corrected according to the formula: wherein n diamond ≈ 2.42
Data Source
Figure 1
Figure 2a
Figure 2b
AI summary
A process operable using a computerized system for grading the clarity of a diamond, the computerized system including an optical image acquisition device, a processor, a pre-trained neural network and an output module operably interconnected together, said process including the steps of (i) acquiring via an optical image acquisition device one or more plurality of axial view images of a diamond with different focus depths, (ii) in a processor, combining the plurality of axial views to form one or more single optical images, wherein the single image comprises in-focus defects from the plurality of axial views and such that out of in-focus defects from the plurality of axial within the diamond are rejected; (iii) in a pre-trained neural network (neural network 120), providing a regressive value associated with the clarity grade of said diamond based on the one or more single images acquired during step (i); wherein the pre-trained neural network has been pre-trained utilising one or more single optical images acquired from a plurality of diamonds each having a preassigned clarity grade assigned and (iv) from an output module, providing a clarity grade to the diamond of (i) by correlating the regression value from (ii) to a clarity grade.