Vehicle Collision Prediction From Image Distance Change
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current object collision prediction methods in intelligent driver assistance systems require significant computing power and storage space to identify target types and calculate distances, leading to inefficiencies and errors, especially when data is not stored in databases.
Innovation Solution
A method that involves capturing two images at different moments to measure distance changes between an object and a to-be-detected target, predicting collisions based on these changes without identifying target types, and calculating relative velocities to estimate collision times, thereby reducing computational load and storage needs.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If object type identification and database storage are used for collision prediction, then measurement precision is improved, but device complexity and computing power requirements increase
Solution Approach 1:
The patent extracts only the essential information needed for collision prediction (distance and velocity changes) from the complex object identification process. Instead of identifying full object types and storing them in databases, the system extracts purely geometric and kinematic parameters directly from image sequences, eliminating the need for complex classification systems while maintaining prediction accuracy
Solution Approach 2:
The patent segments the collision prediction task into independent measurement components: distance calculation from image coordinates, velocity derivation from distance changes over time, and collision determination from velocity analysis. This segmentation allows each component to be processed separately using simple image processing techniques rather than requiring a unified complex identification system
2Measurement precision
If object type identification and database storage are used for collision prediction, then measurement precision is improved, but computing power consumption increases
Solution Approach 1:
The patent uses simple, computationally inexpensive image processing operations that can be performed rapidly on standard hardware. Instead of requiring powerful processors for complex object recognition, the system uses basic coordinate extraction and arithmetic operations that consume minimal computing power while achieving the same predictive function
Solution Approach 2:
The patent replaces the mechanical/computational system of object identification and database lookup with a direct mathematical measurement approach. By using image coordinate geometry and temporal changes to derive distance and velocity, the system substitutes complex information processing with straightforward calculations that require minimal computational resources
Data Source
Figure 1A
Figure 1B
Figure 2
AI summary
This application provides a collision detection method and an apparatus. An image shot by a photographing unit may be used to predict whether a collision with a to-be-detected target occurs. In a current collision prediction method, a type of the to-be-detected target needs to be determined first based on the image shot by the photographing unit, which requires consuming of a large amount of computing power. In the collision prediction method provided in this application, a change trend of a distance between the to-be-detected target and a vehicle in which the apparatus is located may be determined based on distances between the to-be-detected target and the vehicle in which the apparatus is located in images shot at different moments, to predict whether a collision occurs between the to-be-detected target and the vehicle in which the apparatus is located. This method can improve efficiency in performing collision prediction, and reduce energy consumption for performing collision prediction.