Araştırma Makalesi
BibTex RIS Kaynak Göster

Handwritten Digit Recognition with Boundary Curve and Skeleton Features

Yıl 2025, Cilt: 15 Sayı: 1, 577 - 591, 15.03.2025
https://doi.org/10.31466/kfbd.1621840

Öz

In this study, an application related to handwritten digit recognition is introduced. The feature data used for the recognition of handwritten digits are extracted from the boundary curve and skeletal structure of the digit shape. These two feature sets show complementary properties in recognizing the formal features of digits. First, handwritten digit images are parsed in the background and converted into black-white binary images, and in the following stage, 40x40 image matrix data are obtained by normalization process. Based on these matrix data, the boundary curve and skeletal forms of the handwritten character are created. 300 handwritten digit samples are used in the study. A data set is produced by using different features extracted from the boundary curve and skeletal forms. The classification of handwritten digits is provided by using an Artificial Neural Network (ANN) trained with the obtained data set. With the use of a limited number of samples, the recognition success is 96.7%.

Kaynakça

  • Kimura F.,(1991), "Handwritten Numerical Recognition Based on Multiple Algorithms", Pattern Recognition, Vol.29. No.7.
  • Duerr B., Haettich W., Tropf H., Winkler G., (1980),"A combition of statistical and synctactical pattern recognition applied to classification of unconstrained handwritten numerals", Pattern Recognition, Vol.12, Issue 3, 1980, Pages 189-199
  • Cheng D., (1998), "Recognition of Handwritten Digits Based on Contour Information", Pergamon Pattern Recognition,Vol.29. No.7.
  • Basak J., (1995), "A Connectionist System for Learning and Recognition of Structures: Application to Handwritten Characters", Pergamon, Neural Networks,Vol.8, No.4.
  • Leung C.H., Sze L., 1997, "Feature Selection in the Recognition of Handwritten Chinese Character1, Engn Applic. Artif. Intell. Vol.10,No.2
  • Amin A.,(1996), "Handprinted Arabic Character Recognition System Using an Artificial Network", (1996),Pergamon, Pattern Recognition,Vol.29. No.4.
  • Trier D., (1996), "Feature Extraction Methods for Character Recognition", Pergamon, Pattern Recognition,Vol.29. No.4.
  • Trahanias P.E., 1992, "Binary Shape Recognition Using the Morphological Skeleton Transform", Pattern Recognition, Vol.25, No.11.
  • Lam L., Lee S.-W., Suen C.Y., (1992),"Thining methodologies- acomprehensive survey", IEEE Transactions on Pattern Analysis and Machine Intelligence, Volume: 14, Issue: 9.
  • Cheng-Yuan L., "Handprinted Character Recognition Based on Spatial Topology Distance Measurement", (1996), IEEE Trans. On Pattern An. and Mac. Intell. ,Vol.18, No.9.
  • Huang K., Yan H., 1997, "Off-Line Signature Verification Based on Geometric Feature Extraction and Neural Network Classification", Pattern Recognition, Vol.30, no.1.
  • Xu-Yao Zhang, Yoshua Bengio, Cheng-Lin Liu, "Online and Offline Handwritten Chinese Character Recognition: A Comprehensive Study and New Benchmark", (2016), Computer Vision and Pattern Recognition.
  • Jayasekara S., Rajasegaran J., Jathushan R., Seneviratne S., Rodrigo R., "Textcaps: Handwritten character recognition with very small datasets", (2019), IEEE Winter Conference on Applications of Computer Vision (WACV)
  • De Stefano C., Fontanella F., Freca A., "A ranking-based feature selection approach for handwritten character recognition", (2019), Pattern Recognition Letters, Vol.121, Pages 77-86.
  • WenY., Ke W., Sheng H., "Improved Handwritten Numeral Recognition on MNIST Dataset with YOLO and LSTM", (2022), IEEE International Conference on Universal Village (UV).

Sınır Eğrisi ve İskelet Özellikleri ile El Yazısı Rakam Tanıma

Yıl 2025, Cilt: 15 Sayı: 1, 577 - 591, 15.03.2025
https://doi.org/10.31466/kfbd.1621840

Öz

Bu çalışmada el yazısı rakam tanımaya ilişkin bir uygulama tanıtılmaktadır. El yazısı rakamların tanınması amacıyla kullanılan öznitelik verileri, rakam şeklinin sınır eğrisi ve iskelet yapısından çıkartılmaktadır. Bu iki öznitelik seti, rakamlara ait biçimsel özellikleri tanımada tamamlayıcı özellikler göstermektedir. El yazısı rakam görselleri öncelikle arka zeminde ayrıştırılarak siyah-beyaz ikili görüntüye dönüştürülmüş ve izleyen aşamada normalizasyon işlemi ile 40x40 boyutlarında görüntü matris verileri elde edilmiştir. Bu matris verileri temel alınarak el yazısı karakterin sınır eğrisi ve iskelet formları oluşturulmuştur. Çalışmada 300 adet el yazısı rakam örnekleri kullanılmıştır. Sınır eğrisi ve iskelet formlarından çıkartılan farklı öznitelikler kullanılarak bir veri seti üretilmiştir. Elde edilen veri seti ile eğitilen bir Yapay Sinir Ağı (YSA) kullanılarak el yazısı rakamların sınıflandırılması sağlanmıştır. Sınırlı sayıda örnek kullanımı ile birlikte tanıma başarısı %96.7 olarak elde edilmiştir.

Etik Beyan

Yapılan çalışmada araştırma ve yayın etiğine uyulmuştur.

Kaynakça

  • Kimura F.,(1991), "Handwritten Numerical Recognition Based on Multiple Algorithms", Pattern Recognition, Vol.29. No.7.
  • Duerr B., Haettich W., Tropf H., Winkler G., (1980),"A combition of statistical and synctactical pattern recognition applied to classification of unconstrained handwritten numerals", Pattern Recognition, Vol.12, Issue 3, 1980, Pages 189-199
  • Cheng D., (1998), "Recognition of Handwritten Digits Based on Contour Information", Pergamon Pattern Recognition,Vol.29. No.7.
  • Basak J., (1995), "A Connectionist System for Learning and Recognition of Structures: Application to Handwritten Characters", Pergamon, Neural Networks,Vol.8, No.4.
  • Leung C.H., Sze L., 1997, "Feature Selection in the Recognition of Handwritten Chinese Character1, Engn Applic. Artif. Intell. Vol.10,No.2
  • Amin A.,(1996), "Handprinted Arabic Character Recognition System Using an Artificial Network", (1996),Pergamon, Pattern Recognition,Vol.29. No.4.
  • Trier D., (1996), "Feature Extraction Methods for Character Recognition", Pergamon, Pattern Recognition,Vol.29. No.4.
  • Trahanias P.E., 1992, "Binary Shape Recognition Using the Morphological Skeleton Transform", Pattern Recognition, Vol.25, No.11.
  • Lam L., Lee S.-W., Suen C.Y., (1992),"Thining methodologies- acomprehensive survey", IEEE Transactions on Pattern Analysis and Machine Intelligence, Volume: 14, Issue: 9.
  • Cheng-Yuan L., "Handprinted Character Recognition Based on Spatial Topology Distance Measurement", (1996), IEEE Trans. On Pattern An. and Mac. Intell. ,Vol.18, No.9.
  • Huang K., Yan H., 1997, "Off-Line Signature Verification Based on Geometric Feature Extraction and Neural Network Classification", Pattern Recognition, Vol.30, no.1.
  • Xu-Yao Zhang, Yoshua Bengio, Cheng-Lin Liu, "Online and Offline Handwritten Chinese Character Recognition: A Comprehensive Study and New Benchmark", (2016), Computer Vision and Pattern Recognition.
  • Jayasekara S., Rajasegaran J., Jathushan R., Seneviratne S., Rodrigo R., "Textcaps: Handwritten character recognition with very small datasets", (2019), IEEE Winter Conference on Applications of Computer Vision (WACV)
  • De Stefano C., Fontanella F., Freca A., "A ranking-based feature selection approach for handwritten character recognition", (2019), Pattern Recognition Letters, Vol.121, Pages 77-86.
  • WenY., Ke W., Sheng H., "Improved Handwritten Numeral Recognition on MNIST Dataset with YOLO and LSTM", (2022), IEEE International Conference on Universal Village (UV).
Toplam 15 adet kaynakça vardır.

Ayrıntılar

Birincil Dil Türkçe
Konular Bilgisayar Yazılımı
Bölüm Makaleler
Yazarlar

Hasan Hüseyin Çelik 0000-0003-2885-0501

İhsan Gök 0009-0007-3633-9099

Yayımlanma Tarihi 15 Mart 2025
Gönderilme Tarihi 17 Ocak 2025
Kabul Tarihi 14 Mart 2025
Yayımlandığı Sayı Yıl 2025 Cilt: 15 Sayı: 1

Kaynak Göster

APA Çelik, H. H., & Gök, İ. (2025). Sınır Eğrisi ve İskelet Özellikleri ile El Yazısı Rakam Tanıma. Karadeniz Fen Bilimleri Dergisi, 15(1), 577-591. https://doi.org/10.31466/kfbd.1621840