Identyfikatory
Warianty tytułu
Handwritten digit pattern recognition based on education robot
Języki publikacji
Abstrakty
Niniejsza praca prezentuje zaimplementowanie systemu rozpoznającego ręcznie pisane wzorce cyfrowe z użyciem mobilnego układu edukacyjnego LEGO Mindstorms NXT. Został on wybrany ze względu na prostotę w konstrukcji i równocześnie możliwość złożonego programowania. Zbudowany w ramach projektu robot skanujący znaki pisma ręcznego spełnił założenia początkowe. Wyniki zaimplementowanego algorytmu rozpoznającego również pokryły się z oczekiwaniami - system osiągnął skuteczność na poziomie 100% w warunkach idealnych. We względnie utrudnionych warunkach skuteczność spadła do 91%.
Pattern recognition can be classified depending on the data source, the way data is read, processed and on the implementation of the recognition itself [9]. This paper presents a method of pattern recognition identifying handwritten Arabic numbers. The data is collected by a Lego Mindstorms NXT 2.0 mobile robot using a color sensor. Usually, the input data are gathered by high-precision equipment [2,5], and or have an additional multi-sensor subsystem [1]. Very successive recognition approaches are based on neural networks [3, 4,6] additional supported by statistic [8]. Unfortunately, all these methods require powerful calculations. The environment data read by such a simple educational robot contains many drawbacks: noises, relative stabile confidence etc. The solution we propose solves to some extent the problem using a minimal hardware equipment (Fig. 4) and undemanding computation effort. The built recognition system is divided into two parts. The first part presents the data set collection - the hand-written digits scanning (Fig.1) and the data initial processing. The second one consists of primary and secondary classification (Figs. 2 and 3). The algorithm is based on the undirected graph model [10]. The results of the conducted experiments are very interesting (Tabs. 1 and 2). This encourages further exploration of implementation of the well-known and new recognition methods on minimal hardware.
Wydawca
Czasopismo
Rocznik
Tom
Strony
812--814
Opis fizyczny
Bibliogr. 11 poz., rys., tab.
Twórcy
autor
- Politechnika Białostocka, ul. Wiejska 45A, 15-351 Białystok
autor
- Politechnika Białostocka, ul. wiejska 45A, 15-351 Białystok
Bibliografia
- [1] Pfister M., Behnke S., Rojas M.: Recognition of Handwritten ZIP Codes in a Real-World Non-Standard-Letter Sorting System, Applied Intelligence 12 pp. 95–115, Kluwer Acad. Publ. 2000.
- [2] Sesmero M. P., Alonso-Weber J. M., Gutie´rrez G., Ledezma A., Sanchis A.: A new artificial neural network ensemble based on feature selection and class recoding, Neural Comp&Applic, 21, pp. 771-783, Springer-Verlag London, 2010.
- [3] Hiroki Kurashige, Hideyuki Cˆateau: A Method to Construct Visual Recognition Algorithms on the Basis of Neural Activity Data, ICONIP 2011, LNCS 7064, pp. 485-484, Springer-Verlag Berlin Heidelberg 2011.
- [4] Liu Benyong Zhang Jing: An adaptively trained kernel-based nonlinear representation for handwritten digit classification, Journal of Electronics (China), Vol. 23, No. 3, 2006 pp. 379-384.
- [5] Lei Huang, Zhen Li: Feature-based image registration using the shape context, Intern. Journal of Remote Sensing, Vol. 31, No. 8, 2010, pp. 2169- 2177.
- [6] Manabu Kotani, Seiichi Ozaw: Feature Extraction Using Independent Components of Each Category, Neural Processing Letter, Vol. 22, 2005, pp. 113-124.
- [7] Hairong Lv and Wenyuan Wang: Handwritten Digit Recognition with Kernel-Based LVQ Classifier in Input Space, LNCS 3497, pp. 203–208, 2005, Springer-Verlag Berlin Heidelberg 2005.
- [8] Marisa Morita1,2, Robert Sabourin1,2,3, Fl´avio Bortolozzi3, Ching Y. Suen: Segmentation and recognition of handwritten dates: an HMM-MLP hybrid approach, IJDAR (2004) 6: pp. 248–262.
- [9] Rahman1 A. F. R., Fairhurst M. C.: Multiple classifier decision combination strategies for character recognition: A review, IJDAR (2003) 5: pp. 166–194.
- [10] Gajer M.: Systemy optycznego rozpoznawania znaków pisma. Pomiary Automatyka Robotyka 4/2008, s. 21-25.
- [11] Steven S. S.: The Algorithm Design Manual. Springer-Verlag, London 2012.
Typ dokumentu
Bibliografia
Identyfikator YADDA
bwmeta1.element.baztech-2f45b174-c6ab-433b-a5ac-5cc63b043216