Abstract:A technique and an algorithm of digital surface image processing are proposed to increase the validity of real-time detection of small size defects. The algorithm is implemented in the MATLAB programming environment. The technique is based on segmentation of the high-frequency component of surface texture because small size defects are especially pronounced in this component. The high-frequency component, in particular roughness, is extracted by means of wavelet transform for frequency components separation and homomorphic filtration for compensation of low-frequency distortion caused by nonuniform illumination of test surface. Segmentation of the high-frequency texture component consists in formation of a binary image using the texture descriptors derived from the gray-level co-occurrence matrix as the segmentation threshold. The proposed technique and algorithm are approved in applications to defect detection for a simulated surface, for real ground surface of hardened steel, and for surfaces of carbon fiber reinforced plastic composite. Extraction efficiency of the high-frequency component of surface texture is shown. It is found that texture descriptors, "contrast' and "energy," can be applied as segmentation thresholds for defect extraction/determination on the ground (anisotropic) surface while segmentation of an image of a plastic composite (isotropic) surface is effective just with "energy" as a threshold. The proposed technique can be applied for simultaneously real-time monitoring the surface texture and detecting the small size defect in machine vision systems during production and operation of tribosystems.

Sensor Defects in Digital Image Forensic

Image Forensics Technology Using Abnormity of Local Hue for Blur Detection

Detecting Image Tampering Using Feature Fusion

A Novel Defect Detection System for Complex Freeform Surface Structures

Defective Kernel Detection Using A Linear Colour Ccd

On the Security and Applicability of Fragile Camera Fingerprints

Digital Image Forgery Detection Based on Lens and Sensor Aberration

Capturing high-resolution digital images for use in forensic document examination

Digital Image Tamper Detection Techniques - A Comprehensive Study

Improving Image Interpretation in Digital Forensics

Defect Detection Approaches Based on Simulated Reference Image

Recent Advances in Passive Digital Image Security Forensics: A Brief Review

Learning to Detect Multiple Photographic Defects

FixPix: Fixing Bad Pixels using Deep Learning

Segment Anything in Defect Detection

Technique of Real-Time Detection of Technical Surface Defects

The Adobe Hidden Feature and its Impact on Sensor Attribution

A survey on digital image forensic methods based on blind forgery detection

An Analysis of Optical Contributions to a Photo-Sensor's Ballistic Fingerprints

A Stress Test for Robustness of Photo Response Nonuniformity (Camera Sensor Fingerprint) Identification on Smartphones

Research on Surface Defect Detection Technology Based on the Zonal and Time-Sharing Computational Imaging