Real-Time Medical Article Recognition via Local Feature Segmentation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing techniques fail to recognize medical articles such as devices, instruments, or pharmaceutical products in real-time within captured videos.

Innovation Solution

An information processing system that utilizes local feature extraction and recognition, including a local feature storing unit, a feature generating unit, and a recognizing unit to identify medical articles by comparing feature vectors across images, allowing for real-time recognition and management in hospital and pharmacy settings.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Measurement precision

If traditional image recognition techniques (SIFT transform, k-means clustering, Support Vector Machines) are used to identify medical articles, then recognition accuracy can be achieved, but real-time recognition cannot be achieved due to computational complexity

Engineering Contradiction:
Improverecognition accuracyVSAvoidrecognition speed
Core Design Contradiction:
Measurement precisionVSProductivity

Solution Approach 1:

The patent segments the medical article image into multiple blocks and extracts local features from each block independently. This segmentation approach reduces the computational complexity of processing the entire image at once, enabling faster feature extraction while maintaining recognition accuracy through comprehensive local feature coverage.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The patent extracts only the essential local features (edges, corners, blobs) from the image using specialized detectors, rather than processing all image data. This selective extraction of critical features reduces computational load significantly while preserving the most discriminative information needed for accurate medical article recognition.

Inventive Principle:
Principle #2Taking out (Extraction)

2Measurement precision

If comprehensive feature extraction is performed to ensure accurate recognition, then recognition accuracy improves, but processing time increases preventing real-time operation

Engineering Contradiction:
Improverecognition accuracyVSAvoidprocessing time
Core Design Contradiction:
Measurement precisionVSLoss of time

Solution Approach 1:

The patent applies different feature extraction strategies to different local regions of the image. Each block is processed with appropriate feature detectors based on its characteristics, allowing efficient processing of simple regions while applying more comprehensive analysis only where needed, thus balancing accuracy and speed.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent performs preliminary feature extraction in advance, creating a feature database of medical articles before actual recognition occurs. This pre-processing step allows the system to quickly compare extracted features against known patterns during real-time operation, significantly reducing processing time while maintaining accurate recognition.

Inventive Principle:
Principle #10Preliminary action

Data Source

PatentEP2811459B1Information processing system, information processing method, information processing device, and control method and control program therefor, and communication terminal, and control method and control program therefor
Publication Date: 2020.02.19 NEC CORP
  • EP2811459B1 patent drawingFigure 1
  • EP2811459B1 patent drawingFigure 2
  • EP2811459B1 patent drawingFigure 3

AI summary

Provided is a technique for recognizing, in real time, a medical article such as a medical device, a medical instrument, or a pharmaceutical product in an image in a video. A medical article and m-number of first local features which are respectively constituted by a feature vector of 1 dimension to i dimensions of m-number of feature points in an image of the medical article are stored in association with each other, n-number of feature points are extracted from an image in a captured video, n-number of second local features respectively constituted by a feature vector of 1 dimension to j dimensions are generated, a smaller number of dimensions among the number of dimensions i and the number of dimensions j is selected, and an existence of the medical article in the image in the video is recognized when it is determined that a prescribed ratio or more of the m-number of first local features up to the selected number of dimensions corresponds to the n-number of second local features up to the selected number of dimensions.