MLP를 이용한 3D 필름 패턴 영상의 품질 검사
Quality Inspection of 3D Film Pattern Images Using MLP
0
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
3D 패턴 필름은 빛의 양과 각도에 따라 2D 패턴 필름을 3D로 보이게 하는 필름이다. 최근에는 기계학습을 이용한 제품 품질을 검사하는 연구가 많이 발표되고 있다. 하지만 데이터의 양이 적고 유사한 영상들이 대부분일 때 기계학습을 사용하게 되면 과적합이 발생하게 되어 정확도가 낮아지게 된다. 또한 3D 패턴 필름은 윤곽선이 흐리기 때문에 검출하기가 어렵다는 한계가 있다. 본 연구에서는 MLP(Multiplayer Perceptron) 모델을 이용하여 3D 패턴 필름의 품질을 검사하는 방법을 제안한다. 이는 3D 패턴 필름의 영상 히스토그램에서 특정 지점의 높이로부터 폭을 구하여 MLP 모델의 입력 값으로 사용한다. 그리고 나서 최적의 하이퍼 파라미터를 찾은 후에 제품의 품질을 검사한다. 제안한 방법은 적은 수의 영상 자료에 대하여 MLP를 사용하여 과적합의 가능성을 줄이고 높은 정확도를 달성할 수 있다는 장점이 있다. 실험 결과, 제안된 방법의 정확도는 98.809%로써 다른 모델에 비해 우수한 성능을 보였다.
Abstract
3D pattern film is a type of film that, depending on the amount and angle of light, creates the illusion of a 2D pattern film appearing in 3D. Recently, there have been numerous publications on inspecting product quality using machine learning. However, when dealing with small datasets consisting mostly of similar images, the use of machine learning can lead to overfitting and reduced accuracy. Additionally, one limitation of 3D pattern films is the difficulty in detecting contours due to their blurred nature. In this study, we propose a method for inspecting the quality of 3D pattern films using the MLP(Multiplayer Perceptron) model. This involves extracting the width from a specific point to the height in the image histogram of the 3D pattern film, which is then utilized as input for the MLP model. Subsequently, after finding the optimal hyperparameters, the product's quality is evaluated. The proposed approach offers the advantage of using MLP to mitigate the risk of overfitting with a limited number of image data points, thereby achieving high accuracy. Experimental results showed that the proposed method exhibited superior performance with an accuracy of 98.809% compared to other models.
참고문헌 (14)
- B. S. Rao, "Dynamic histogram equalization for contrast enhancement for digital images", Applied Soft Computing, 89, pp.106114~ (2020)
- J. Lee, "3D film image inspection based on the width of optimized height of histogram", Journal of the Institute of Convergence Signal Processing, 23(2), pp.107~114 (2022)
- J. N. Mlyahilu, "Morphological geodesic active contour algorithm for the segmentation of the histogram-equalized welding bead image edges", IET Image Processing, 16(10), pp.2680~2696 (2022)
- A. G. Medeiros, "An automatic left ventricle segmentation on echocardiogram exams via morphological geodesic active contour with adaptive external energy", Journal of Artificial Intelligence and Systems, 1(1), pp.77~95 (2019)
- J. Mlyahilu, "Classification of 3D Film Patterns with Deep Learning", Journal of Computer and Communications, 7(12), pp.158~165 (2019)
- A. Salman, "Leaf Classification and Identification using Canny Edge Detector and SVM Classifier", 2017International Conference on Inventive Systems and Control (ICISC), pp.1~4 (2017)
- S. Mascarenhas, "A comparison between VGG16, VGG19 and ResNet50 architecture frameworks for Image Classification", 2021International conference on disruptive technologies for multi-disciplinary research and applications (CENTCON), pp.96~99 (2021)
- Y. H. Qu, "Paper defects classification based on VGG16 and transfer learning", Journal of Korea TAPPI, 53(2), pp.5~14 (2021)
- D. Theckedath, "Detecting affect states using VGG16, ResNet50 and SE-ResNet50 networks", SN Computer Science, 1(2), pp.1~7 (2020)
- K. S. Kumar, "ResNet50 : Automated Fabric Defect Detection and Classification based on a Deep Learning Approach", Journal of Propulsion Technology, 44(4), pp.2156~2165 (2023)
- Y. Wang, "Generalizing from a few examples : A survey on few-shot learning", ACM computing surveys, 53(3), pp.1~34 (2020)
- Y. Cao, "An effective industrial defect classification method under the few-shot setting via two-stream training", Optics and Lasers in Engineering, 161, pp.107294~ (2023)
- J. Mlyahilu, "A Fast Fourier Transform with Brute Force Algorithm for Detection and Localization of White Points on 3D Film Pattern Images", Journal of Imaging Science and Technology, 66, pp.030506~ (2021)
- D. R. I. M. Setiadi, "PSNR vs. SSIM : Imperceptibility quality assessment for image steganography", Multimedia Tools and Applications, 80(6), pp.8423~8444 (2021)