Railway crossing risk area detection using linear regression and terrain drop compensation techniques

Area of Science:

• Computer Vision
• Railway Engineering
• Object Detection

Background:

• Railway accidents frequently occur at crossings.
• Detecting people and objects in these high-risk zones is crucial for accident prevention.

Purpose of the Study:

• To develop and validate an effective method for detecting railway crossing risk areas.
• To enhance safety by identifying potential hazards at railway crossings.

Main Methods:

• Utilized terrain drop compensation (TDC) to address railway crossing concavity.
• Employed linear regression for object position and length prediction from image data.
• Introduced a novel calculating local maximum Y-coordinate object points (CLMYOP) strategy for ground point acquisition.
• Applied image preprocessing to reduce noise and improve detection accuracy.

Main Results:

• The proposed scheme effectively detects objects and humans in railway crossing risk areas.
• Experimental results confirm the accuracy and reliability of the developed detection methods.
• Image preprocessing significantly enhanced the overall object detection performance.

Conclusions:

• The integrated approach, combining TDC, linear regression, and CLMYOP, provides an effective solution for railway crossing risk detection.
• The developed system offers a corrective and efficient method for improving railway safety.
• This research contributes to the prevention of railway accidents through enhanced situational awareness at crossings.