Millimeter Wave Gender Recognition via Region Segmentation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional gender recognition methods in millimeter wave imaging systems suffer from low calculation efficiency and recognition accuracy due to the differences between millimeter wave and visible light imaging mechanisms, with millimeter wave images having low gray levels and high noise, and relying on single feature information for classification being insufficient.
Innovation Solution
A human body gender automatic recognition method and apparatus that involves acquiring millimeter wave grayscale images, determining gender part region positions, extracting and normalizing region sub-images, extracting shape and gray variance features, and integrating classification results from multiple classifiers to improve recognition accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Device complexity
If traditional visible light image processing methods are applied to millimeter wave images, then the recognition process can be simplified, but the recognition accuracy deteriorates due to the fundamental differences in imaging mechanisms
Solution Approach 1:
The patent changes the parameter space from visible light image parameters to millimeter wave image parameters. It extracts features specifically suitable for millimeter wave images (such as chest region features, crotch region features, and gray variance information) rather than directly applying visible light image processing algorithms, thereby adapting the recognition system to the fundamental differences in imaging mechanisms while maintaining accuracy
Solution Approach 2:
The patent segments the millimeter wave image into different body regions (chest region and crotch region) and extracts features from each region separately. This segmentation allows the system to capture gender-specific characteristics in different body parts, improving recognition accuracy by focusing on discriminative features rather than processing the entire image as a whole
2Productivity
If single feature information (such as gray variance information) is used for classification, then the calculation efficiency can be improved, but the recognition accuracy deteriorates due to insufficient feature representation
Solution Approach 1:
The patent merges multiple feature types (shape features, chest region features, crotch region features, and gray variance information) into a comprehensive feature set. By combining these diverse features, the system achieves both high recognition accuracy through comprehensive feature representation and maintained calculation efficiency through efficient feature extraction and integration methods
Solution Approach 2:
The patent transitions from single-dimensional feature extraction to multi-dimensional feature space. It extracts features from multiple body regions and multiple feature types simultaneously, creating a high-dimensional feature representation that captures comprehensive gender characteristics while maintaining computational efficiency through optimized feature processing
3Measurement precision
If multiple feature information and multiple classifiers are integrated, then the recognition accuracy can be improved, but the calculation complexity increases
Solution Approach 1:
The patent performs preliminary feature extraction and normalization before classification. By pre-processing the millimeter wave images to extract and normalize features from different body regions, the system reduces the complexity of subsequent classification operations, allowing multiple classifiers to work more efficiently on prepared features rather than raw images
Solution Approach 2:
The patent introduces feature normalization as an intermediary step between image extraction and classification. This normalization process serves as a mediator that standardizes features from different regions and types, making them comparable and reducing the complexity of integrating multiple classifiers by providing a unified feature representation space
Data Source
AI summary
A human body gender automatic recognition method includes: acquiring a current millimeter wave grayscale image, and determining gender part region positions of a human body in the millimeter wave grayscale image according to a pre-set body proportion; extracting a region sub-image corresponding to the gender part region position; performing dimension normalization on the region sub-image to obtain a normalized region sub-image; performing feature information extraction on the normalized region sub-image to obtain feature information about the normalized region sub-image; recognizing the millimeter wave grayscale image by means of each pre-set classifier, and respectively outputting results; and integrating the output results to obtain a classification recognition result of the millimeter wave grayscale image. The method has relatively high recognition rate and calculation efficiency and solves the problem of a privacy protection and detection method in a millimeter wave security inspection system.