Parallel ISP Fault Detection Using Split Raw Image Overlap
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Image signal processors (ISPs) in image processing systems can generate inaccurate images due to faults, leading to issues in advanced driver assistance systems (ADAS) functions.
Innovation Solution
A method is provided for processing raw image signals using multiple ISPs by splitting the signal into partial signals, processing them independently, and comparing overlapping regions to detect faults using a fault detection circuit.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If multiple ISPs are used to process raw image signals, then processing capability and productivity are improved, but the complexity of the system increases and fault detection becomes more difficult
Solution Approach 1:
The patent divides the image signal processing task into separate processing paths, with each ISP processing a specific region or portion of the raw image signal. This segmentation allows multiple ISPs to operate in parallel, improving processing capability while maintaining manageable complexity through modular architecture.
Solution Approach 2:
The patent introduces a fault detection circuit as an intermediary component that receives outputs from multiple ISPs and compares them to detect faults. This intermediary mechanism enables fault detection without requiring complex interconnections between all ISPs, thus managing system complexity while maintaining high productivity.
2Reliability
If multiple ISPs process the same raw image signal, then processing redundancy is improved for fault detection, but the amount of data to be processed increases
Solution Approach 1:
The patent segments the raw image signal into different portions that are distributed to multiple ISPs. Each ISP processes only its assigned portion, reducing the data volume each individual processor handles while maintaining overall processing redundancy for fault detection.
Solution Approach 2:
The patent employs partial processing where each ISP handles only a subset of the image signal data. This partial action approach provides sufficient redundancy for fault detection without requiring each ISP to process the entire image signal, thus controlling data volume management.
3Measurement precision
If fault detection is performed by comparing images from multiple ISPs, then detection accuracy is improved, but processing time increases
Solution Approach 1:
The patent performs preliminary processing of the raw image signal by distributing it to multiple ISPs in parallel before fault detection is needed. This preliminary action allows fault detection to occur without requiring sequential processing, thus maintaining high detection accuracy while minimizing processing time through concurrent operations.
Solution Approach 2:
The patent enables continuous processing where multiple ISPs operate simultaneously on different portions of the image signal. This continuity of useful action allows fault detection to proceed without interrupting the processing flow, maintaining both high accuracy and efficient processing time through overlapping operations.
Data Source
Figure 1~2
Figure 3
Figure 4
AI summary
A method of determining an image signal processor (ISP) fault performed by an image processing device is disclosed. The method includes generating a first partial raw image signal and a second partial raw image such that a split circuit includes each blending region, generating, by a first ISP, a first partial image based on the first partial raw image signal, generating, by a second ISP, a second partial image based on a second partial raw image signal, and determining, based on the first partial image and the second partial image, whether a fault occurs in an ISP group including the first ISP and the second ISP.