What is Handwriting Recognition?
Handwriting recognition (HandWriting Recognition) refers to the process of informatizing the trajectory generated when writing on a handwriting device into text.
- jHWR handwriting recognition system is based on advanced large-capacity dictionary technology. It is an efficient and convenient means to provide real-time and accurate handwriting recognition services to anyone at any time, any place, and is in line with the dynamic update and personalization of the information age. Query requirements.
- JHWR architecture
- Online handwriting recognition refers to transforming the ordered trajectory information generated when writing on a handwriting device into
- Offline handwriting recognition involves automatically converting text in an image into a character code that a computer can use. Offline handwriting recognition is more difficult because different people have different writing styles. Offline handwriting recognition is mainly used for printed text recognition.
- Techniques to reduce identification errors
- It is often used to narrow down the recognition range. For example, the postal code contains only numbers from 1 to 9. Recognizing such numbers can reduce the possibility of errors.
- The main technologies:
- Specify a specific character range;
- Take advantage of the unique characteristics of characters [1]
- General process
- Online handwriting recognition can be broken down into several general steps:
- Pre-processing
- Feature extraction
- Separate characters.
- The purpose of preprocessing is to discard irrelevant input data, which can reduce negative effects. This involves speed and accuracy. It usually consists of image binarization, normalization, sampling, smoothing, and denoising preprocessing.
- (1) eJHWR technology, using the syntactic structure self-learning algorithm and multi-core fusion technology based on feature statistical algorithm. Has the following characteristics:
- 1. High recognition rate.
- 2. Fast recognition speed.
- 3. No stroke order restrictions.
- 4. The size of the data dictionary can be expanded.
- (Two) eJHWR support environment
- OS: WinCE, Nucleus, Embedded Linux, Symbian, Palm Os, HOPEN, pSOS, UCOS, etc.
- CPU: DragonBall (Motorola 68000) EZ (16M) and VZ (33M), MIPS, SH3, SH4, ARM7, StrongARM, ARM9, Xscale etc
- (3) eJHWR can bring the following main functions:
- 1. Text input Instead of frequent keystrokes entered by the keyboard, frequent input mode switching is avoided when text symbols are mixed, and text that is uncertain in pronunciation can also be written as usual.
- 2. Quick Positioning-Use the stylus as a mouse, fast positioning, much easier than keyboard operation.
- 3. Quick query When the menu level is too deep or the exact classification of the query information is not known, handwriting input is called, which avoids frequent and blank keyboard operations to achieve quick query [2] .