1. Online identification of the font in the picture
At present, the Internet "word Passenger network", "Font", "myfonts", "Whatfontis" and other sites to provide image font recognition services, the site in the identification accuracy of different. We might as well look for our own identification channels in these sites.
The use of each site is basically similar. Here's how to quickly identify fonts in a picture, as an example of the word guest network. Prepare a picture with a special font (such as a picture with "computer enthusiasts" written above).
In the browser address bar to access http://www.fontke.com/, click on the upper right "font recognition" button (or direct access to the Literacy Web site http://www.fontke.com/likefont/), click "Local Picture", select the computer picture, Click on the "Open" button to upload the picture (see Figure 1).
The text split results appear below the text settings (see Figure 2). In some split results, you may have a text that is split into multiple parts, and then drag the picture to the target picture box to combine the parts into a complete text. Then enter the exact text below each text and click on the "Start Recognition" button.
Small hint: What to do if the text splits up missing word?
The Braille service can support images of complex colors, and automatically process thresholds. If individual tints or darker text or parts are missing, you can adjust the threshold value progress bar until the text in the picture box is displayed correctly. If the picture frame shows white and black background, click the "Reverse" button to treat the image in reverse.
Wait a moment, the word net will search for matching fonts, drag the original to the appropriate location to be identified (see Figure 3). After you find the exact font, click on the link to the font to download the trial or purchase the Font business license. The preview text for the font in the search results is the text entered in the text settings, but you can always modify the preview text, resize the font, set the foreground color, or the background colors.
Tips: Improve the accuracy of complex image recognition
If the text of the uploaded picture appears to be sticky, the image should be preprocessed. Text settings Only enter text that differs significantly from other fonts, which can greatly improve the accuracy of recognition, because many fonts may have a phenomenon in which some text is basically close to other text.
2. Comparative selection of literacy websites
We use different types of picture cases (common English, Special English, Common Chinese, special Chinese, complex Chinese), identified by 4 different websites, and the results are listed in the table (table). You can refer to the results of this table, according to their identified font type, choose a more appropriate image font recognition site platform.
From the test results, "The word Guest Network" and "font" support in English, Chinese, Japanese, Korean and other fonts; MyFonts and Whatfontis only support English fonts. Judging from the accuracy of the test, "word passenger network" for several test pictures can be accurately identified; Font "For most of the test pictures are also identified, a few test pictures to identify the approximate font; MyFonts mainly support English commercial fonts, the free English fonts can not be accurately identified; Whatfontis can accurately identify English fonts, but in order to prioritize the most accurate precedence in front. From the support of the image of the complex program, "The word guest NET" can support more complex composition of the picture, "Font", myfonts, Whatfontis can support a relatively simple picture composition. In terms of functionality, the "Word Guest network" supports reverse phase, threshold, and other image post-processing functions, the search results of the font preview user experience is very good; Font "Support the computer automatic spelling function, but the site ads too much, seriously affect the user experience;" MyFonts supports image OCR function, but not quite accurate for some unconventional text OCR; Whatfontis in the identification results does not support the original image of the pull, affecting the user experience of comparison.
After more than a variety of fonts compared to the test, the author's overall feeling is that the use of "Word Guest network" image recognition font services, the process is simpler, the support of the language more widely, while the recognition accuracy is also relatively higher. The other 3 kinds of website service also has its director, can decide the choice according to the actual situation.