គិតទៅ គិតចុះ គិតឡើង ខ្ញុំកំសត់ខ្លួនដល់ហើយ ព្រោះខ្ញុំមិនសូវចេះភាសារ ទាំងអង់គ្លេស បារាំង បូករួមទាំងភាសារកំណើតផងដែរ។ ខ្ញុំសរសេរមិនសូវត្រូវទេ។ ហើយរឿងចំបងនោះ គឺខ្ញុំមិនសូវយកចិត្តទុកដាក់រៀនសូត្រ តាំងតែពីដើមមក។ ដល់ពេលនេះទើបគិតថាត្រូវស្តាយ តែធ្វើមិច បើហួសពេលហើយ។ ចឹងហើយ មានតែប្រឹងៗទៅ តាំងត្រដាបត្រដួស។ ខាងក្រោមនេះជាអត្ថបទដែលខ្ញុំសាកសរសេរមួយ មិនចង់អោយគេមើលទេខ្មាស់គេ។ តែរឿងខ្មាស់ណឹងទុកម៉ាកាដុបសិនចុះ បើមិនខំរៀនសរសេរទេ មិនដឹងជាពេលណាខ្ញុំចេះសរសេរ។ បើអ្នកណាគេបានអាន អត្ថបទខាងក្រោយណឹង ខ្ញុំសុំសួរតែមួយទេ គឺយល់ថាខ្ញុំចង់សរសេរអំពីអ្វីរឺអត់? ព្រោះខ្លាចតែសរសេរហើយគេមើលមិនយល់ផង។ ខ្ញុំមិនសួដល់កំហុសវេយ្យារណ៏ រឺ ស្ទីសរសេរបែបក្មេងទើបតែរៀនដូចខ្ញុំសរសេរនេះទេ។
Since 1950, the Optical Character Recognition (OCR) had been developed for a commercial OCR in the US. Then this method was using in an IBM application call “Optical Reader” in 1960. In 1970 the commercial OCRs were beginning pervasive in Japan. Then the handwriting recognition becomes also an interesting subject in research domain including handwriting isolate characters and handwriting isolate word. There are a lot of documents writing for example administrative document, meeting note are produced everyday and need to be digitized. Some researches in this domain can be noticed for example …..
There are two types of the handwriting documents, off-line and on-line documents. The off-line documents are the document was digitized by scanning or capturing the paper version and converting to the image version. The on-line document are documents which was written based on digital Ink technology that allow to capture the writing stroke in real time and store its coordinate X, Y, and or times and pen pressure. This online document was started in research domain since 1960 and it was be interested in 1980 because of the development of the technology. Nowadays, there are many kinds of digital Ink such as Personal Digital Assistants (PDAs), electronic tablet and other technology with pen base.
There are two mains approaches to solve this problem known as the global approaches and analytic approaches. The global approaches suppose each word image or signal as an element or class and try to recognize with its entry characteristics. These approaches does not looking to recognize the characters containing in each word. The analytic approaches were proposed with the converse objective of global approaches. These approaches try to recognize the characters containing in each word. The advantage of analytic approach is to make indecently between the learning phases with the changing of lexicon. With these approaches, only the alphabets of each language should be trained to the system and normally adapted for the alphabets languages. With this main advantage, the analytic approaches become interesting approaches in these last few years. We can notice some recent research …. But the main difficulty in these approaches is the segmentation and selection the characters sequence of the word independently. To avoid this problem, our research proposed a bi-character approach to ensure that the sequence of characters selected are correct. And this method is adapted for lexicon base and non lexicon base.