Partial Recognition Process

Through a process called partial recognition, the recognition engine uses some or all of the user-entered stroke data to determine the resulting Unicode match. Windows CE uses partial recognition for glyphs in character sets with complex multi-stroke stylus input, such as Japanese, Chinese, and Korean. You can set up the HwxSetPartial function to accomplish partial recognition of glyphs similarly to how you set it up for character recognition. The exception is that your application calls the recognition engine multiple times for each glyph recognition. Each time a stroke is processed through the UI, you instruct the recognition engine to process by using the current amount of data.

Though not an absolute requirement, multi-threaded operation is the most efficient means for performing partial recognition. While one thread gathers character input, another can process data. When the two threads communicate with each other, the thread that performs character input can control the thread that performs the character recognition. This gives your application the capability to produce partial results while the user is still writing. The HwxSetAbort function enables the gathering thread to tell the processing thread to stop, saving processing time by not using an invalid result. In this basic process, HwxSetAbort passes the recognition engine an HRC and a pointer to an address that contains the number of strokes that are currently input by the user. If the contents of the address do not match the number of strokes that the recognition engine is trying to process, the recognition engine stops.