Passenger Counting Depth Distance Calculation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing passenger counting systems face errors in face detection, often incorrectly counting multiple individuals at close distances as a single person, leading to inaccurate passenger counts.
Innovation Solution
A passenger counting system that calculates the movement amount of a vehicle and determines the depth distance of passenger faces using a pinhole camera model, allowing for accurate differentiation between individuals based on depth and traveling direction thresholds.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If face detection is performed to count passengers, then the passenger number can be obtained, but erroneous detection occurs when multiple persons are at close distance
Solution Approach 1:
The patent transitions from two-dimensional image-based face detection to three-dimensional spatial analysis by calculating depth distances of detected faces. By computing the depth direction distances of multiple detected faces and comparing them against threshold values, the system can distinguish between faces that are close in the image plane but at different depths, thereby resolving erroneous detections when multiple persons are at close distance
Solution Approach 2:
The patent introduces depth direction distance as an additional parameter to the traditional face detection process. By calculating and utilizing the depth distance of each detected face, the system transforms the detection criteria from solely based on image coordinates to a multi-parameter approach including spatial depth information, improving the reliability of passenger counting
Data Source
AI summary
The erroneous detection determination system acquires a movement amount of a vehicle. The erroneous detection determination system calculates, based on the movement amount of the vehicle, distances in a depth direction of objects detected as faces of passengers of the vehicle from an image of the vehicle. The erroneous detection determination system determines whether the detected objects include an object that has been erroneously detected, based on the distances in the depth direction of the objects.