Multi-Angle Illumination Target Detection Using Gray Scale Differences
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing target detection methods in image processing are complex and require significant computational resources, making them inefficient for simple and effective target detection in target scenes.
Innovation Solution
A target detection method that uses images acquired by an image acquiring device with light sources at different illuminating angles to determine the presence of a target based on gray scale differences, employing a direct and oblique incidence light source setup and normalized cross-correlation algorithms to calculate similarity values and identify target characteristics.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If hybrid Gaussian algorithm is used for target detection, then detection accuracy is improved, but computational complexity increases significantly
Solution Approach 1:
The patent segments the target detection process into two distinct phases: an offline training phase where the Gaussian model is pre-built using hybrid Gaussian algorithm, and an online detection phase where only simple threshold comparison is performed. This segmentation transfers computational complexity from the detection phase to the training phase, resolving the contradiction between detection accuracy and real-time computational complexity.
Solution Approach 2:
The patent performs preliminary action by pre-processing images to extract feature values before building the Gaussian model. The offline training phase pre-computes the Gaussian distribution parameters (mean and variance) for each feature, so that during online detection, only simple threshold comparisons are needed. This preliminary computation resolves the technical contradiction by moving complex calculations away from the real-time detection process.
2Reliability
If hybrid Gaussian algorithm with continuous model updating is used, then detection reliability is improved, but detection speed decreases
Solution Approach 1:
The patent segments the detection system into an offline model training component and an online detection component. The Gaussian model is trained offline with continuous updating to improve reliability, while the online detection uses the pre-trained model for fast inference. This temporal segmentation resolves the contradiction between detection reliability and detection speed.
Solution Approach 2:
The patent creates a simplified copy of the detection process that uses the pre-trained Gaussian model parameters (mean and variance) for threshold-based detection. Instead of continuously updating the full Gaussian model during online detection, the system uses a simplified threshold comparison based on pre-computed parameters, maintaining reliability while improving speed.
3Measurement precision
If multiple images with different illuminating angles are used, then target detection accuracy is improved, but device complexity increases
Solution Approach 1:
The patent applies local quality by using different illuminating angles for different types of targets. The system selects specific light sources (direct incidence, oblique incidence from different directions) based on the detection needs, rather than using all possible configurations simultaneously. This selective approach improves detection accuracy for various target types while controlling device complexity.
Solution Approach 2:
The patent makes the light source system multi-functional by using the same image acquiring device and light sources for both training phase data collection and detection phase operation. The system universally handles different target types (small targets, large targets, targets with different reflectivity) using the same hardware configuration with varying illumination angles, reducing overall device complexity.
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
This method simplifies target detection by leveraging gray scale differences to determine target presence and characteristics, enabling effective detection of both large and small targets with reduced computational complexity.
Implementation Method 1
obtaining a plurality of images acquired by an image acquiring device when a region to be detected is illuminated by light sources with different illuminating angles
Implementation Method 2
determining whether a target exists in the region to be detected based on gray scale differences among the obtained images
Data Source
AI summary
The embodiments of the present application provides a target detection method, which comprises: obtaining a plurality of images acquired by an image acquiring device as images to be detected when a region to be detected is illuminated by light sources with different illuminating angles, wherein the different illuminating angles correspond to the different images to be detected (S101); and determining whether a target exists in the region to be detected based on gray scale differences between the obtained images to be detected with illuminating by the light sources of with different illuminating angles (S102). The method is applied to achieve a simple and effective detection of the target existing in the region to be detected.