I'm experimenting with Abbyy finereader and "recognizing words with spaces" as described into user guide.
From my tests, it seems that a phase inserted in the user dictionary does not always produce better recognizion of the written text (hand written).
I wrote two italian words and applyed an user dictionary containing these single two words and also the phrase given by the two words separated by space. I added the space into the italian charset too.
The restul was that i can correctly recognize the single words when i draw small areas around each of them. But if i draw a large area surrounding both words i get i wrong recognition.
I expected that an user dictionary with few words shoud give a strong hint to te engine to detect the right text. It seems instead no great difference from applying the standard italian dictionary, which produced almost the same wrong recognition of the whole phrase.
Any advice to try to get better results?
Thanks, Michele
コメント
1件のコメント
Hi Michele,
The first recommendation would be to increase the Weight parameter of the user dictionary you add. By default the DictionaryDescription::Weight of any dictionary is 100, but by setting it to a higher value, you can make FineReader Engine use words from this dictionary.
Also please note, that RecognizerParams::OneWordPerLine option might have a significant influence.
On an unrelated issue, please note that you should specifically set that you recognize handprinted text. ABBYY FineReader Engine does not support automatic recognition of text as handprinted. The recognition quality will decrease significantly if you do not set it manually. For more information on that topic, please refer to Help → Guided Tour → Advanced Techniques → Recognizing Handprinted Texts.
If this will not be helpful, please send us examples of the files you are dealing with and full description of your scenario to SDK_Support@abbyy.com.
サインインしてコメントを残してください。