Light Field Camera Defect Pixel Detection via Ray Directional Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional real-time defect detection and correction methods for light field cameras face challenges in distinguishing defect pixels from object edges due to the inclusion of ray directional information and the large number of photoelectric conversion elements, leading to increased processing time and memory resource usage.
Innovation Solution
An image processing apparatus that reconstructs the picked-up image based on ray directional information to generate a reconstruction image, allowing for real-time defect pixel detection and correction, with detection parameters adjusted according to refocus information to improve precision and reduce memory usage.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If real-time defect detection is performed on picked-up images with ray directional information, then defect pixels can be detected, but processing time increases and memory resource usage increases
Solution Approach 1:
The patent performs defect detection on the picked-up image before the refocus reconstruction process. By detecting defects in advance on the raw picked-up image data, the system avoids the need to process large amounts of reconstruction data, thereby reducing processing time while maintaining defect detection capability
Solution Approach 2:
The patent extracts and detects defect pixels from the picked-up image separately from the main refocus reconstruction process. This allows defect detection to be performed on a subset of data rather than the entire reconstruction dataset, reducing memory resource usage and processing time
2Reliability
If defect detection is performed on picked-up images with ray directional information, then defect pixels can be detected, but memory resource usage increases
Solution Approach 1:
The patent performs defect detection on the picked-up image before the refocus reconstruction process. By detecting defects in advance on the raw picked-up image data, the system avoids the need to process large amounts of reconstruction data, thereby reducing processing time while maintaining defect detection capability
Solution Approach 2:
The patent extracts and detects defect pixels from the picked-up image separately from the main refocus reconstruction process. This allows defect detection to be performed on a subset of data rather than the entire reconstruction dataset, reducing memory resource usage and processing time
3Reliability
If conventional defect detection methods are used on light field camera images, then defect pixels can be detected, but precision decreases due to difficulty in distinguishing defect pixels from object edges
Solution Approach 1:
The patent adjusts the detection threshold dynamically based on local image characteristics and refocus information. By making the threshold adaptive rather than fixed, the system can distinguish defect pixels from object edges more accurately in different regions of the image
Solution Approach 2:
The patent uses refocus information as feedback to improve defect detection precision. The refocus reconstruction results are used to verify and refine defect pixel identification, allowing the system to distinguish true defects from edge artifacts more effectively
Data Source
AI summary
An image processing apparatus obtains a picked-up image which is output from an image pickup element which includes a microlens array for obtaining the picked-up image including ray directional information of an object image formed by a photographing optical system, generates a reconstruction image by reconstructing the obtained picked-up image on the basis of the ray directional information, and detects a defect pixel of the generated reconstruction image.