Image Evaluation Reliability Filtering for Accurate Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional image classification methods struggle to accurately evaluate shooting states due to unstable acquisition of auto-focus and camera shake evaluation values, particularly in scenes with low contrast or background focus errors, leading to improper evaluation of image quality.
Innovation Solution
An image processing apparatus and method that acquires images, evaluates their quality based on reliable evaluation values, and records a rating result, excluding values with low reliability to ensure accurate image classification across various shooting scenes.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional image classification methods use auto-focus evaluation values and camera shake evaluation values, then image classification can be performed, but the evaluation values are unstable and cannot be acquired reliably in certain shooting conditions
Solution Approach 1:
The patent changes the parameter being measured from raw evaluation values (auto-focus, camera shake) to reliability values that indicate the trustworthiness of these evaluation values. By introducing a reliability parameter, the system can adapt to different shooting conditions by weighting or discarding evaluation values based on their reliability, thus resolving the contradiction between reliability and adaptability across various scenes.
2Measurement precision
If auto-focus evaluation is performed on subjects with low contrast or background focus, then focus evaluation can be conducted, but the evaluation values become incorrect or unstable
Solution Approach 1:
The patent introduces reliability values as an intermediary that mediates between the raw evaluation values and the final image classification. Instead of directly using potentially incorrect evaluation values from difficult shooting conditions, the system first assesses the reliability of these values and then uses them appropriately, thus resolving the measurement precision issue in challenging scenes.
3Productivity
If all evaluation values are used for image classification, then comprehensive evaluation is achieved, but unreliable evaluation values degrade the overall classification accuracy
Solution Approach 1:
The patent applies local quality by treating different evaluation values differently based on their reliability. Instead of uniformly using all evaluation values, the system assigns different weights or validity statuses to each evaluation value based on its reliability assessment. This allows the system to maintain high classification accuracy by excluding unreliable local evaluations while still utilizing reliable ones, thus resolving the contradiction between productivity and measurement precision.
Data Source
AI summary
An image processing apparatus comprises an image acquisition unit configured to acquire an image, an evaluation value acquisition unit configured to acquire an evaluation value of the image, a reliability acquisition unit configured to acquire reliability of the evaluation value, an evaluation unit configured to evaluate the image based on the evaluation value; and a recording unit configured to add a rating result of the image to the image and record the rating result, wherein the evaluation unit performs rating in accordance with the evaluation value excluding an evaluation value the reliability of which is relatively low.