Localization of Steel Ball in Impact Perforation Images by Neural Network

Ogawa, Takehiko; Kanada, Hajime; Kasano, Hideaki

doi:10.9746/sicetr1965.39.789

Cited by 4 publications

(3 citation statements)

References 2 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The multilayered neural network model to distinguish a steel ball in the impact perforation image has been studied [3] [4]. However, it is difficult to prepare the appropriate training data set, because of the variance of the fragment shapes.…”

Section: Introductionmentioning

confidence: 99%

Impact perforation image processing using a self-organizing map

Okubo

Ogawa

Kanada

2007

SICE Annual Conference 2007

View full text Add to dashboard Cite

A characteristic evaluation method from an image of impact perforation process by a high speed photograph system has been studied in the material engineering field. In this method, a steel ball is shot into a material specimen to evaluate characteristics by observing the ball and fragments of the specimen. However, it is often difficult to find a steel ball because of the dispersion of fragments. In this study, we propose to use the self-organizing map to distinguish a steel ball, background and fragments.

show abstract

Section: Introductionmentioning

confidence: 99%

Impact perforation image processing using a self-organizing map

Okubo

Ogawa

Kanada

2007

SICE Annual Conference 2007

View full text Add to dashboard Cite

show abstract

“…We propose the use of neural network to classify them in the image with degradation by fragments of the specimen. We show the image classification of the images degraded by the scattered material fragments accurately by using the neural network (Ogawa et al, 2003). The successive images of the impact perforation test with actual composite materials; polycarbonate (PC), polymethyl methacrylate (PMMA) and alumina (Al 2 O 3 ) specimens are shown in Fig.2.…”

Section: Impact Perforation Testmentioning

confidence: 99%

Image Clustering and Evaluation on Impact Perforation Test by Self-Organizing Map

Ogawa¹

2010

Self-Organizing Maps

Self Cite

View full text Add to dashboard Cite

“…We have already proposed to apply the multilayer neural network for the recognition of the steel ball in the impact perforation image. We have used the three-layered feed-forward neural network which estimated the steel ball location in the impact perforation image [3]. However, the miss-recognition of the steel ball was often seen because of the influence on the scattered fragments of the specimen.…”

mentioning

confidence: 99%

Impact Perforation Image Processing Using a Neural Network

Ogawa

Tanaka

Kanada

et al. 2006

2006 SICE-ICASE International Joint Conference

Self Cite

View full text Add to dashboard Cite

The evaluation of material's characteristics from the impact perforation images has been studied in the material engineering fields. In this method, the steel ball is shot into the material specimen, and the characteristic ofthe material is estimated from the steel ball's behavior. However, the observation of steel ball's behavior is often difficult because of the scattered fragments of the specimen. We have proposed to use the neural network to estimate the steel ball position in the impact perforation image. However, the miss-recognition of the steel ball was often seen because of the influence on the scattered fragments of the specimen. In this study, the preprocessing of the image with the high-pass filter is introduced to improve the performance of the recognition of the steel ball. We examine two types of filters using the Hanning window and the Blackman window.Keywords: impact perforation images, steel ball, neural network, high-pass filter. INTRODUCTIONIt is important to estimate the perforation limit velocity and residual velocity of the material in the design of the structure that collides with the flying object. Therefore, the characteristic evaluation from the impact perforation image by the super high-speed photograph system has been studied in the material engineering [1] [2]. In this method, a steel ball is shot into the material specimen and is localized by the sequential photograph. Then, the characteristic of the material is estimated from the steel ball location and the specimen's behavior. However, the observation of steel ball's behavior might be difficult in the specimen such as ceramics because of scattering the fragments.Neural networks are often used as a method of the image recognition. When the neural network is used for the image recognition, we expect the correct recognition from an imperfect image because of the robustness of the neural network. We have already proposed to apply the multilayer neural network for the recognition of the steel ball in the impact perforation image. We have used the three-layered feed-forward neural network which estimated the steel ball location in the impact perforation image [3]. However, the miss-recognition of the steel ball was often seen because of the influence on the scattered fragments of the specimen.On the other hands, the high-pass filters are often used for edge emphasis in the image processing. Moreover, the high-pass filters have an effect to make the background of the image uniform, since they cut the low frequency.In this study, we introduce the preprocessing of the image with the high-pass filter to the neural network to improve the performance of the recognition of the steel ball. We examine two types of filters using the Hanning window and the Blackman window. We use the impact perforation image of the polymethyl methacrylate (PMMA) specimen which contains a lot of scattered fragments.

show abstract

Localization of Steel Ball in Impact Perforation Images by Neural Network

Cited by 4 publications

References 2 publications

Impact perforation image processing using a self-organizing map

Impact perforation image processing using a self-organizing map

Image Clustering and Evaluation on Impact Perforation Test by Self-Organizing Map

Impact Perforation Image Processing Using a Neural Network

Contact Info

Product

Resources

About