Video Crossline and Direction Detection Using Pixel-Line Frames
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing video surveillance systems face high computational intensity when detecting objects crossing a crossline and determining the direction of crossing due to extensive image processing requirements.
Innovation Solution
A method involving the extraction of two or more parallel lines of pixels from each video frame corresponding to the crossline, constructing a combined image frame, and using a machine learning model to detect crossings and directions, reducing computational load by processing only relevant pixel data.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If object detection and tracking is performed on the entire video sequence using image processing, then crossing detection and direction detection are achieved, but computational intensity becomes excessively high
Solution Approach 1:
The video frames are segmented to extract only the specific lines corresponding to the crossline region. Instead of processing entire frames, the system divides the image space and processes only the relevant horizontal lines where crossings occur, significantly reducing computational load while maintaining detection accuracy.
Solution Approach 2:
The patent extracts only the necessary pixel data from video frames - specifically the lines corresponding to the crossline and a limited region around it. This extraction approach removes unnecessary computational processing of areas that do not contain crossing events, directly addressing the contradiction between accurate detection and computational efficiency.
2Reliability
If full image processing is applied to detect all objects in the scene, then comprehensive object detection is achieved, but processing time increases significantly
Solution Approach 1:
The system applies different processing quality levels to different regions of the video frames. High-quality processing is applied only to the crossline region where crossings occur, while the rest of the frame receives minimal or no processing. This local quality approach maintains reliability for the critical function while reducing overall processing time.
Data Source
AI summary
A device and method for detecting objects crossing a crossline, and the direction, captured by a video camera are described. A sequence of video image frames of the scene is captured, and a combined image frame is created by extracting two or more lines of pixels of each image frame and arranging them adjacent to each other, wherein the lines of pixels of each video image frame are parallel and correspond to the crossline in the scene. The combined image frame is sent to a machine learning model that detects a combined image frame representing an object crossing a crossline and a direction the object crosses during capturing of the sequence of video image frames. A detection of any object crossing the crossline and a direction of crossing is received from the machine learning model.