Object Recognition Device Resolution Upscaling
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional object recognition techniques fail to accurately recognize objects in images with different resolutions, leading to reduced recognition precision due to loss of high-frequency components when aligning resolutions by reducing or increasing image quality.
Innovation Solution
An object recognition system that increases the resolution of low-resolution images using a pre-learned acquisition process and optimizes neural network parameters to extract feature vectors from the increased-resolution images, allowing for precise object recognition by minimizing differences between feature vectors of the same and different labeled images.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the resolution of the query image is increased using conventional methods, then the resolution is aligned with reference images, but high-frequency components are lost and recognition precision deteriorates
Solution Approach 1:
The system performs resolution increase processing on the query image before feature extraction and object recognition. By pre-processing the low-resolution query image to increase its resolution to match reference images, the system preserves high-frequency components during subsequent processing steps, thereby maintaining recognition precision without information loss
Solution Approach 2:
The system changes the resolution parameter of the query image through learned upscaling methods. Instead of simply reducing reference image resolution or using conventional upscaling that loses high-frequency information, the system applies parameter changes through neural network-based resolution increase processing that preserves critical image details
2Measurement precision
If the resolution of the reference image is reduced to match the query image, then resolution alignment is achieved, but detailed information is lost and recognition precision deteriorates
Solution Approach 1:
Instead of reducing the reference image resolution to match the query image (conventional approach), the system inverts the approach by increasing the query image resolution to match the reference images. This inversion preserves the detailed information in reference images while achieving resolution alignment through intelligent upscaling of the query image
3Adaptability or versatility
If different resolution images are used for query and reference, then processing flexibility is maintained, but feature vector consistency deteriorates leading to wrong object retrieval
Solution Approach 1:
The system performs resolution increase processing on the query image before feature extraction to ensure that both query and reference images have consistent resolution. This preliminary resolution alignment ensures that feature vectors extracted from both images are comparable, maintaining consistency while preserving the system's flexibility to process images of various original resolutions
Data Source
AI summary
An object included in a low-resolution image can be recognized with a high degree of precision. An acquisition unit acquires, from a query image, an increased-resolution image, which is acquired by increasing the resolution of the query image, by performing pre-learned acquisition processing for increasing the resolution of an image. A feature extraction unit, using the increased-resolution image as input, extracts a feature vector of the increased-resolution image by performing pre-learned extraction processing for extracting a feature vector of an image. A recognition unit recognizes an object captured on the increased-resolution image on the basis of the feature vector of the increased-resolution image and outputs the recognized object as the object captured on the query image.