Quality Predictor for Image Retrieval Recognition Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
There is a gap between printed media and electronic media, with no mechanism for publishers to easily migrate printed content into electronic form with augmented content, and existing image recognition technologies struggle with low-quality images from mobile devices, requiring improved speed and accuracy in image recognition and handling multiple recognition algorithms for mixed media environments.
Innovation Solution
A Mixed Media Reality (MMR) system that includes mobile devices, an MMR gateway, an MMR matching unit, and an MMR publisher, utilizing a quality predictor to filter images and route queries to appropriate recognition units, and combining recognition results to provide a single result, while also enabling registration of images and content for newspaper publishing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If image recognition is performed on low-quality images from mobile devices, then image retrieval can be enabled, but recognition accuracy deteriorates
Solution Approach 1:
The system performs preliminary actions by pre-processing images during registration to generate multiple versions with different quality levels and applying multiple recognition algorithms in advance. This allows the system to handle low-quality query images by comparing them against pre-processed reference images of varying qualities, thereby improving recognition accuracy without requiring perfect input image quality.
Solution Approach 2:
The system changes parameters by adjusting image quality parameters and applying different recognition algorithms based on image characteristics. The quality predictor analyzes image parameters to determine appropriate processing paths, and the system dynamically selects recognition algorithms and processing parameters based on the quality metrics of the input images, enabling accurate recognition across varying image qualities.
2Reliability
If multiple recognition algorithms are used to improve recognition accuracy, then recognition robustness is improved, but device complexity increases
Solution Approach 1:
The quality predictor serves as an intermediary that manages the complexity of multiple recognition algorithms. It analyzes input image characteristics and automatically selects the most appropriate recognition algorithm or combination of algorithms, thereby maintaining high recognition robustness while simplifying the system's operational complexity by providing a centralized decision-making layer.
Solution Approach 2:
The system dynamically selects and adjusts recognition algorithms based on real-time image quality assessment. Rather than rigidly applying all algorithms to all images, the system adapts its processing approach based on the specific characteristics of each input image, optimizing the balance between recognition robustness and computational complexity through dynamic algorithm selection.
3Productivity
If image quality filtering is applied to improve recognition speed, then processing efficiency is improved, but loss of information occurs
Solution Approach 1:
The system applies partial filtering by selectively processing images based on quality thresholds rather than uniformly filtering all images. The quality predictor identifies images that meet minimum quality criteria and routes them through optimized recognition paths, while preserving more detailed processing for borderline cases. This partial application of filtering maintains recognition speed for clear images while preventing information loss for images that benefit from more thorough analysis.
Data Source
AI summary
A MMR system for newspaper publishing comprises a plurality of mobile devices, an MMR gateway, an MMR matching unit and an MMR publisher. The MMR matching unit receives an image query from the MMR gateway and sends it to one or more of the recognition units to identify a result including a document, the page and the location on the page. The MMR system also includes a quality predictor as a plug-in installed on the mobile device to filter images before they are included as part of a retrieval request or as part of the MMR matching unit. The quality predictor comprises an input for receiving recognition algorithm information, a vector calculator, a score generator and a scoring module. The quality predictor receives as inputs an image query, context information and device parameters, and generates an outputs a recognizability score. The present invention also includes a method for generating robustness features.