Object Recognition Using Density Gradient Voting
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing object recognition systems, particularly for human face recognition, are inefficient due to reliance on simple shape templates that fail to accurately and rapidly identify mask areas, requiring numerous arithmetic operations and being unsuitable for rapid recognition.
Innovation Solution
An object recognition system that converts images into density gradient directional templates and subject images, using a feature analyzer to select reference points, a vote module to calculate candidate centers, and a detection module to determine the mask area based on votes, allowing for accurate and rapid recognition of objects or human faces.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of manufacture
If simple shape templates (circular, ellipse, polygon) are used to extract mask area, then the template selection is simple, but the mask area extraction accuracy deteriorates
Solution Approach 1:
The patent transforms the template from simple geometric shapes to density gradient directional templates with multiple parameters (density gradient values, directional information, reference point coordinates). This allows the template to adapt to various object shapes and sizes while maintaining automated processing, resolving the contradiction between simplicity and accuracy.
Solution Approach 2:
The patent replaces manual template selection with automated image conversion and feature analysis processes. The image converter automatically transforms images into density gradient directional templates, and the feature analyzer automatically identifies reference points, eliminating the need for manual intervention while improving accuracy.
2Measurement precision
If ellipse template with parameter variation is used to match mask area, then the mask area extraction accuracy is improved, but the computational complexity and processing time increase significantly
Solution Approach 1:
The patent performs preliminary image conversion into density gradient directional templates before the actual mask area extraction. This pre-processing step transforms the image data into a format that directly supports accurate mask identification, eliminating the need for iterative parameter variation and correlation analysis during the extraction process.
Solution Approach 2:
The patent extracts key features (reference points with distance and angle parameters) from the density gradient directional template and stores them in a table structure. This extraction allows for rapid lookup and matching during mask area identification, avoiding the need for complex real-time calculations and reducing computational complexity.
3Reliability
If iterative correlation analysis is performed to find optimum ellipse parameters, then the mask area matching is improved, but the recognition speed deteriorates
Solution Approach 1:
The patent creates a density gradient directional template that copies and transforms the essential features of the original image into a standardized format with reference points. This template can be directly matched with subject images without iterative optimization, achieving both reliable matching and fast recognition speeds.
Solution Approach 2:
The patent changes the representation parameters from geometric ellipse parameters to density gradient directional parameters (including distance L and angle α of reference points). This parameter transformation enables direct matching and voting mechanisms that achieve reliable results without iterative computation, significantly improving recognition speed.
Data Source
AI summary
An object recognition system utilizes a mask locator for extracting a mask area (M) from within a subject image in order to recognize the mask area (M) with a subject image. The mask locator (40) refers to density gradient directional images respectively converted from an image size template and the subject image, and analyzes these density gradient directional images particularly with respect to the density gradient directional values (θT, θS) assigned to each pixel in each directional images to obtain parameters linking the image size template to the mask area (M) to be registered with the image size template. The mask locator has a vote module (60) which gives candidates for the center of the mask by varying the parameters and votes on the candidates. Based upon the number of votes for the candidates, a detection module (70) designate one of the candidates having a maximum vote number as the center of the mask area (M), thereby determining the mask area (M) around thus designated candidate within the subject image for exact recognition of the object in the subject image.