Have you ever imagined if the writing of a teacher or a lecturer on the whiteboard in ink markers will be stored in digital form in the computer? Of course, the students will be pleased with this, because they no longer have to bother writing the note down in their notebook. They simply listen to the teacher in the class.
This is possible to materialize later on, because currently the computer vision technology is being developed by experts around the world. One of them, Ir. Linggo Sumarno, who is currently developing the technique to recognize handwritten words, using the combination of artificial neural tissue backpropagation. The technique has been created by the man who graduated from UGM Electrical Engineering Department, using isolated word images in binary format, while the output comes out in word text. "The level of recognition using the technique is 84.79 per cent in identifying 480 words written by 10 writers, and 80.75 percent in identifying 4800 words written by 100 writers," said Linggo Sumarno, the lecturer of Sanata Dharma University Yogyakarta in the promotional exam that gave him a doctorate degree in Electrical Engineering Department, Faculty of Engineering UGM on Thursday (27/7). His promoter is Prof. Adhi Susanto, M.Sc., Ph.D., and co-promoters Prof. Dr. Ir. Sri Widodo, DEA, and Drs. Jazi Eko Istiyanto, M.Sc., Ph.D.
In his research, Sumarno took handwriting samples from 100 writers, who were students in Electrical Engineering and Informatics Technology Sanata Dharma University, Yogyakarta, professors of Electrical Engineering University of Sanata Dharma, as well as parents of students in Adisutjipto Elementary School Yogyakarta.
The man born in Ngawi, 16 December 1968, conveyed that the combination of artificial neural tissues he has been developing consists of five artificial neural tissues with identical structure but different in weight values. Nevertheless, this new neural network is only able to recognize letters, yet to the numbers. "This neural network recognizes words written in Latin, print or the combination of both," said the man who graduated as 1248th doctorate from UGM.
In his dissertation entitled Engineering Introduction to the Study of Hand-Written Words Using Artificial Neural Tissue Backpropagation Combination, Sumarno explained, the method using images of words and images of handwritten letters isolated in binary format, which is the scanning result of data collection sheets in 300 dpi resolution.
Meanwhile, to assist the process of his research, Sumarno used computer vision algorithms that basically require high specs computers. Yet, he thinks, the algorithms developed in this research can also be implemented in embedded processors which are more compact.