我可以将OCR文本控制为有限的字符集吗？ [英] May I control the OCR text to a limited character set?

问题描述

   我正在开发一个利用OCR的应用程序。图片只包含有限的字符集（例如，只有6个大写的英文字符和10个数字，而不是26x2或特殊字符）。我发现
的准确度与预期不符，我想如果我可以输入字符集信息作为优化，那么它可以帮助算法识别文本。该提议附于belwo。

POST vision / v1 / ocr？language = zh-Hans& detectOrientation = true

主持人：api .projectoxford.ai

Ocp-Apim-Subscription-Key：•••••••••••••••••••••••••••••• ••

内容长度：236

内容类型：application / json



{" Url "： http://contoso.com/images/test1.jpg,
<强>" CHARACTERSET" ;:" 0123456789ABCDEF"}

解决方案

Davin，

您是否尝试将语言设为英语？你的图像尺寸是多少？有时，如果图像尺寸太小或太大会严重影响精度。我们建议的输入图像尺寸介于40x40
到2600x2600像素之间。如果可能，您可以向我提供（发送至
[e-mail] ）一张API无法提供的样本图片。准确的结果。

谢谢！

Hi,

    I'm developing an application that leverages OCR. The image only contains limited character set (e.g. only 6 English characters in upper case and 10 digits, not 26x2 or special characters). I found the accuracy is not as expected, and I think if I can input the character set information as optimization, then it may helps the algorithm to recognize the text. The proposal is attached belwo.

POST vision/v1/ocr?language=zh-Hans&detectOrientation =true
Host: api.projectoxford.ai
Ocp-Apim-Subscription-Key: ••••••••••••••••••••••••••••••••
Content-Length: 236
Content-Type: application/json

{ "Url": http://contoso.com/images/test1.jpg, "CharacterSet": "0123456789ABCDEF"}

解决方案

Davin,

Have you tried to set the language to English? And what's your dimensions for your images? Sometimes if the dimensions for an image is too small or too large will significantly impact the accuracy. Our suggested dimensions for input image is between 40x40 to 2600x2600 pixels. If possible, you can provide me (send to [e-mail]) one sample image which the API can't give an accurate result.

Thanks!

这篇关于我可以将OCR文本控制为有限的字符集吗？的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！