Vehicle Light Detection Using Restored Scene Brightness Values
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing vehicle light detection methods struggle to accurately distinguish vehicle lights from other bright objects in varying lighting conditions without fixed camera settings, leading to increased complexity and reduced performance.
Innovation Solution
A device and method that utilizes high dynamic range imaging with auto exposure control, followed by image compression and restoration of scene brightness values, combined with classifiers to identify vehicle lights using scene brightness information and simple structure features.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If traditional blob detection techniques or supervised classification methods are used for vehicle light detection, then detection capability is provided, but the methods require fixed camera settings and fixed image preprocessing, which decreases classifier performance when camera parameters change or preprocessing varies
Solution Approach 1:
The patent applies parameter changes by transforming image brightness values from compressed 8-bit representation to original 12-bit scene brightness values using lookup tables. This allows the detection algorithm to adapt to varying camera settings and preprocessing operations, as the method operates on physical brightness values rather than fixed image representations. The parameter transformation enables the same detection logic to work reliably across different camera configurations without retraining or recalibration.
2Measurement precision
If complex features such as large surroundings or motion information are used to distinguish vehicle lights from reflectors, then detection accuracy is improved, but processing complexity and memory requirements increase significantly
Solution Approach 1:
The patent extracts and utilizes only the essential brightness information from the image, specifically the scene brightness values of candidate spots and their immediate neighborhoods. By focusing solely on brightness characteristics rather than complex spatial patterns or motion data, the method achieves adequate detection accuracy while maintaining low processing complexity. The extraction of brightness values enables differentiation between vehicle lights and reflectors through simple brightness ratio comparisons without requiring analysis of large surrounding areas or temporal motion information.
3Productivity
If image compression is applied to reduce memory usage and processing complexity, then efficiency is improved, but scene brightness information may be lost
Solution Approach 1:
The patent performs preliminary action by storing the relationship between compressed 8-bit grey values and original 12-bit scene brightness values in lookup tables during image compression. This preliminary preservation of brightness information mapping allows the system to maintain scene brightness data even after compression. The lookup tables are pre-computed and stored, enabling rapid retrieval of original brightness values during detection without requiring complex decompression or information reconstruction processes.
Data Source
Figure 1~2
Figure 3~4
Figure 5
AI summary
A device for detecting vehicle lights in an image, the device is configured to receive an image captured by a camera, wherein the image comprises an array of pixels and each pixel of the image has a respective scene brightness value, compress the image to generate a compressed image, wherein the compressed image comprises an array of pixels and each pixel of the compressed image has a respective grey value, calculate for each grey value in the compressed image a corresponding scene brightness value in the image, and detect vehicle lights in the compressed image by using the calculated scene brightness values.