An image processing technique for optimizing industrial defect detection using dehazing algorithms.

Area of Science:

• Computer Vision
• Image Processing
• Industrial Automation

Background:

• Industrial defect detection is crucial but challenged by image degradation from environmental factors like water fog.
• Existing algorithms struggle with blurred images, leading to increased detection difficulty and errors.

Purpose of the Study:

• To develop and validate an industrial defect detection algorithm enhanced with dehazing technology.
• To improve the accuracy and reliability of defect detection in complex, fog-affected industrial environments.

Main Methods:

• An optimized dehazing processing method was applied to industrial images captured in water fog.
• The enhanced images were used with an improved YOLOv8 model for defect detection.
• Performance was evaluated using Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity Index Measure (SSIM).

Main Results:

• The dehazing method achieved an average PSNR of 34.9 dB and an SSIM of 0.951.
• The proposed algorithm demonstrated superior performance compared to Convolutional Neural Networks (CNN) and MADNet models.
• The improved YOLOv8 model showed significantly enhanced defect detection confidence and reduced missed detections.

Conclusions:

• The integrated dehazing and defect detection approach effectively addresses image quality issues caused by water fog.
• This novel method offers a robust solution for industrial defect detection in adverse conditions.
• The technology shows potential for transferability to other applications like search and rescue in smoke-filled environments.