Difference Detection Using Neural Network Encoding Data
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional difference detection devices rely solely on image characteristics output from neural networks without utilizing other informative data, limiting their ability to improve accuracy in detecting differences between images.
Innovation Solution
A difference detection device that uses encoding information, such as encoding amount, intra prediction mode, and transform coefficients, in conjunction with image characteristics, to detect differences between images captured at different times, by associating output values from neural networks and transforming this information into an image format for enhanced detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If only image characteristics from neural networks are used for difference detection, then the device complexity is low, but the measurement precision of difference detection is insufficient
Solution Approach 1:
The patent merges multiple data sources including image characteristics from neural networks, encoding information (encoding amount, intra prediction mode, transform coefficients), and association data into a unified difference detection process. This combination of diverse data types enables more accurate difference detection by leveraging complementary information from each source.
Solution Approach 2:
The patent transitions from using only image characteristics (two-dimensional spatial data) to incorporating encoding information that adds temporal and contextual dimensions. By including data from the encoding process and association information, the detection system operates in multiple dimensions, significantly improving detection accuracy.
2Measurement precision
If encoding information is incorporated into difference detection, then the measurement precision improves, but the loss of information during encoding processing increases
Solution Approach 1:
The patent performs difference detection on encoding information (such as transform coefficients and prediction modes) before the final decoding process. By conducting detection operations at this intermediate stage, the system can extract meaningful differences without waiting for complete decoding, thereby reducing information loss and improving detection accuracy.
3Measurement precision
If manual comparison of captured images is performed, then the measurement precision can be high, but the productivity is low
Solution Approach 1:
The patent implements a feedback mechanism where the neural network processes image data and encoding information, generates difference detection results, and uses association data to refine the detection. This automated feedback loop enables high-speed processing while maintaining high accuracy, eliminating the need for manual comparison.
Data Source
AI summary
A difference detection device includes a difference detection unit configured to, based on association among a first image and a second image captured at different times and illustrating a substantially identical space and encoding information of each of the first image and the second image, detect difference between a third image and a fourth image captured at different times and illustrating a substantially identical space, and the encoding information is information acquired from data including the first image encoded and data including the second image encoded, before inverse transform processing is executed in decoding processing executed on each of the first image and the second image.