Object Detection Using Multi-Threshold Binarization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional recognition object detecting apparatuses face challenges in accurately detecting objects at varying distances due to luminance issues, leading to increased processing load and heat generation, which hinders miniaturization and increases manufacturing costs.
Innovation Solution
A recognition object detecting apparatus with an imaging unit that uses high dynamic range characteristics and multiple threshold values to generate binary images, allowing for accurate detection of objects at different luminance levels without the need for multiple sensitivity adjustments, thereby reducing processing load.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If imaging sensitivity is increased to detect low luminance objects at long distance, then detection accuracy for distant objects improves, but high luminance objects at short distance become saturated
Solution Approach 1:
The luminance range is segmented into multiple ranges, with each threshold value corresponding to a specific luminance range. By dividing the detection task into multiple segments (binary images with different thresholds), the system can simultaneously detect objects across the entire luminance spectrum without saturation or loss of detail.
Solution Approach 2:
The system changes the parameter of threshold values to adapt to different luminance conditions. Instead of using a fixed imaging sensitivity, multiple threshold values are applied to binarize the image data, allowing the detection system to handle both low luminance (distant objects) and high luminance (near objects) effectively.
2Measurement precision
If multiple images are taken with different sensitivity settings to detect objects at various distances, then detection accuracy across all distances improves, but processing load increases
Solution Approach 1:
The system performs preliminary binarization of the single image data using multiple threshold values before object detection. This preliminary action creates multiple binary images that simplify subsequent detection processing, reducing the overall processing load compared to acquiring and processing multiple full-resolution images with different sensitivity settings.
Solution Approach 2:
Instead of creating multiple original images with different sensitivity settings, the system creates multiple binary copies from a single image data set. These binary images serve as simplified representations that retain the necessary information for detection while significantly reducing processing requirements.
3Measurement precision
If multiple images are taken with different sensitivity settings, then detection accuracy improves, but heat generation increases preventing miniaturization
Solution Approach 1:
The processing task is segmented into multiple binarization operations on a single image data set rather than acquiring multiple images. This segmentation approach reduces the total computational workload and associated heat generation, enabling miniaturization while maintaining detection accuracy across different luminance ranges.
4Measurement precision
If multiple images are taken with different sensitivity settings, then detection accuracy improves, but manufacturing cost increases
Solution Approach 1:
The system performs preliminary binarization processing on single image data using multiple threshold values, which simplifies the overall detection system. This approach eliminates the need for complex multi-sensitivity imaging hardware and associated control mechanisms, thereby reducing manufacturing costs while maintaining high detection accuracy.
Data Source
AI summary
A recognition object detecting apparatus is provided which includes an imaging unit which generates image data representing a taken image, and a detection unit which detects a recognition object from the image represented by the image data. The imaging unit has a characteristic in which a relation between luminance and output pixel values varies depending on a luminance range. The detection unit binarizes the output pixel values of the image represented by the image data by using a plurality of threshold values to generate a plurality of binary images, and detects the recognition object based on the plurality of binary images.