Environment Recognition Device for Traffic Light Identification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing environment recognition systems face challenges in accurately identifying target objects, particularly light sources like traffic lights, due to low image resolution and distance, leading to false recognition of similar luminance and size characteristics.
Innovation Solution
An environment recognition device and method that utilize a data retaining unit for luminance associations, a luminance obtaining unit, a specific object provisional determining unit, a grouping unit, a position information obtaining unit, and a specific object determining unit to accurately identify target objects by grouping pixels with similar luminance and relative distance, and determining the number and distribution of target objects to prevent false recognition.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If pixels are grouped based on similar luminance and color characteristics, then the recognition process becomes simpler, but the accuracy of identifying target objects deteriorates due to false recognition
Solution Approach 1:
The patent segments the detection area into multiple regions and groups pixels into target objects based on both luminance similarity and spatial proximity. This segmentation approach prevents false recognition by ensuring that only pixels forming coherent spatial groups are identified as target objects, while maintaining the simplicity of the grouping process.
Solution Approach 2:
The patent transitions from two-dimensional pixel grouping based on luminance to three-dimensional spatial grouping that incorporates depth information. By adding the depth dimension, the system can distinguish between target objects at different distances, preventing false recognition while maintaining operational simplicity through automated multi-dimensional grouping.
2Area of stationary object
If the image resolution is low or the light source is distant, then the detection coverage is improved, but the luminance and size recognition accuracy deteriorates
Solution Approach 1:
The patent merges multiple detection parameters including luminance, color characteristics, spatial position, and depth information into a comprehensive target object identification system. By combining these parameters, the system achieves accurate recognition even when individual parameters like luminance or size are unreliable due to low resolution or distance.
Solution Approach 2:
The patent introduces depth information as an intermediary parameter that mediates between the detection coverage requirement and the recognition accuracy requirement. Depth data allows the system to maintain accurate identification by considering the three-dimensional position of objects, compensating for the unreliability of luminance and size measurements at a distance.
3Productivity
If multiple target objects are grouped together, then the processing efficiency is improved, but the specificity of individual object identification deteriorates
Solution Approach 1:
The patent segments target objects into distinct groups based on spatial relationships and luminance characteristics. By applying segmentation algorithms that consider both proximity and luminance similarity, the system efficiently processes multiple objects while maintaining clear identification of each individual object within the group.
Solution Approach 2:
The patent applies local quality analysis by examining the luminance and spatial characteristics of each target object individually within the group. This allows the system to maintain high specificity for individual object identification while processing multiple objects efficiently through automated local analysis.
Data Source
AI summary
There are provided an environment recognition device and an environment recognition method. An environment recognition device 130 provisionally determines a specific object corresponding to a target portion from a luminance of the target portion, groups target portions of which differences in the width direction, and the height direction are within a first predetermined range and which are provisionally determined to correspond to a same specific object into a target object, sequentially detects, from any target objects, target objects of which differences in the width direction, in the height direction, and in the relative distance are within a second predetermined range, and which are provisionally determined to correspond to a same specific object, thereby specifying a target object group, and determines whether or not the target object group is the specific object according to the number of the target objects in the target object group.