Comparison of Feature Histograms and Co-occurrence Matrix on Analysis of the Light Spectrum Effect for Identification of Surface Roughness with Speckle Images

Abstract

Abstract Speckle imaging is a technique that can be used for digital image analysis. Speckle imaging technology essentially uses a laser to create a pattern of dots through the laser’s interference with an object. The advantage of speckle photography is that it is safe and does not require direct contact with the subject. In this study, speckle imaging techniques were used to analyze the surface roughness of objects. Speckle imaging techniques are used to investigate the influence of the choice of laser spectrum on the determination of the surface roughness of an object. For objects with rough surfaces, it comes in 80 sheets of coarse sandpaper and 2000 sheets of fine sandpaper. This study aims to compare the accuracy of identification results using Histogram Features and Co-occurrence Matrix Features when analyzing the influence of laser wavelength on object spot image identification. The research procedures included identifying rough objects, collecting point image data on rough surfaces using a series of experiments, pre-processing the point images, extracting features using histograms and co-occurrence matrices, and using a naive Bayes classifier. It includes determining surface roughness and performing comparisons between them. The resulting feature histogram and feature appearance matrix. This study shows that the use of Histogram Features and coexistence matrix features yields similar conclusions regarding the influence of the laser beam spectrum in determining the surface roughness of an object. The difference between Histogram Features and Co-occurrence Matrix Features lies in the pattern of detection results. Differences in the detection result patterns may be caused by differences in the characteristics of the Histogram Features and Co-occurrence Matrix Features of each speckle image used in the study. Although the detection result patterns of the Histogram Feature amount and the Co-occurrence Matrix feature amount are different, the accuracy of the detection result is equivalent. The similarity in the accuracy of the detection results may be caused by the similarity in the effectiveness of the two features in detecting the speckle image pattern.