Abnormality Detection Apparatus Using Time-Series Integration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing image analysis techniques for detecting abnormal work frequently result in erroneous detections due to momentary changes unrelated to the work content, such as computer failures or image distortions, leading to inaccurate abnormality assessments.
Innovation Solution
A processing apparatus and method that compute a first degree of abnormality at each image photographing time and a second degree of abnormality by integrating multiple first degrees, with the second degree minimizing the impact of momentary influences, allowing for more accurate detection of abnormal work.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If abnormality detection is performed based on degree of abnormality at each timing, then real-time monitoring capability is improved, but erroneous detection occurs frequently due to momentary factors
Solution Approach 1:
The system performs preliminary actions by continuously collecting degree of abnormality data at each timing and storing it in a time series, before making the final abnormality determination. This allows the system to have real-time monitoring capability while avoiding erroneous detections by using historical data for context.
Solution Approach 2:
The system introduces an intermediary mechanism - a determination unit that acts as a mediator between the real-time degree of abnormality measurements and the final abnormality determination. This intermediary analyzes the time series data and momentary changes to make reliable determinations, reducing erroneous detections while maintaining real-time monitoring.
2Speed
If degree of abnormality is computed at each image photographing time, then responsiveness to abnormal conditions is improved, but false alarms increase due to momentary influences
Solution Approach 1:
The system performs preliminary actions by continuously collecting degree of abnormality data at each timing and storing it in a time series, before making the final abnormality determination. This allows the system to have real-time monitoring capability while avoiding erroneous detections by using historical data for context.
Solution Approach 2:
The system performs periodic measurements of the degree of abnormality at each image photographing time, creating a time series of data points. By analyzing this periodic data collection pattern, the system can distinguish between momentary fluctuations and sustained abnormal conditions, reducing false alarms while maintaining responsiveness.
3Measurement precision
If multiple first degrees of abnormality are integrated to compute second degree, then detection accuracy is improved, but computational complexity increases
Solution Approach 1:
The system performs preliminary actions by organizing degree of abnormality data into time series and pre-processing this data before the final determination. This preliminary organization simplifies the subsequent integration process, making it more efficient despite needing to process multiple data points.
Solution Approach 2:
The determination unit serves as an intermediary that manages the integration of multiple first degrees of abnormality. It handles the computational complexity by systematically processing the time series data and applying the integration logic, thereby improving detection accuracy while managing computational requirements in a structured manner.
Data Source
AI summary
In order to detect an abnormal work with high accuracy by image analysis, the present invention provides a processing apparatus 10 including a first computation unit 11 that computes a first degree of abnormality being a degree of abnormality of a work at each image photographing time, based on a processing result of each of a plurality of images acquired by photographing a state of a first work being performed for a predetermined time; and a second computation unit 12 that computes a second degree of abnormality being a degree of abnormality of the first work, based on a plurality of the first degrees of abnormality computed in association with each of the plurality of images.