Tesseract與ios的集成

最近接觸了一個關於圖文識別的項目，項目組決定使用Tesseract,大概查了一下可是對於ios的支持貌似不是很好，如下是爲Tesseract包裝過的oc版本。

 

原文連接：https://github.com/ldiqual/tesseract-ios

 

Tesseract for iOS

tesseract-ios is not actively maintained anymore. I encourage you to use gali8's Tesseract-OCR-iOS instead.

About

Tesseract-ios is an Objective-C wrapper for Tesseract OCR.

This project couldn't exist without the Ângelo Suzuki's blog post. A lot of code came from his article.

Requirements

• iOS SDK 6.0, iOS 5.0+ (there is no support for armv6)
• Tesseract and Leptonica libraries from the tesseract-ios-lib repo.

Installation

• Clone this repo from your project folder.
• Download an appropriate tesseract language trained data from the following website:https://code.google.com/p/tesseract-ocr/downloads/list and put it in your project folder
• You should have the following folder structure:

• Add tesseract-ios as a group, and tessdata by reference to your project:

• Go to your project settings, and ensure that C++ Standard Library => libstdc++:

Usage

Here is the default workflow to extract text from an image:

• Instantiate Tesseract with data path and language
• Set variables (character set, …)
• Set the image to analyze
• Start recognition
• Get recognized text
• Clear

Code Sample

#import "Tesseract.h"

Tesseract* tesseract = [[Tesseract alloc] initWithDataPath:@"tessdata" language:@"eng"];
[tesseract setVariableValue:@"0123456789" forKey:@"tessedit_char_whitelist"];
[tesseract setImage:[UIImage imageNamed:@"image_sample.jpg"]];
[tesseract recognize];

NSLog(@"%@", [tesseract recognizedText]);
[tesseract clear];

Method reference

-initWithDataPath:language:

- (id)initWithDataPath:(NSString *)dataPath language:(NSString *)language

Initialize a new Tesseract instance.

• dataPath: a relative path from the application bundle to the .traineddata files. You can find these files from the tesseract downloads section.
• language: language used for recognition. Ex: eng. Tesseract will search for a eng.traineddatafile in the dataPath directory.

Returns nil if instanciation failed.

-setVariableValue:forKey:

- (void)setVariableValue:(NSString *)value forKey:(NSString *)key

Set Tesseract variable key to value. See http://www.sk-spell.sk.cx/tesseract-ocr-en-variables for a complete (but not up-to-date) list.

For instance, use tessedit_char_whitelist to restrict characters to a specific set.

-setImage:

- (void)setImage:(UIImage *)image

Set the image to recognize.

-setLanguage:

- (BOOL)setLanguage:(NSString *)language

Override the language defined with -initWithDataPath:language:.

-recognize

- (BOOL)recognize

Start text recognition. You might want to launch this process in background with NSObject's -performSelectorInBackground:withObject:.

-recognizedText

- (NSString *)recognizedText

Get the text extracted from the image.

-clear

- (void) clear

Clears Tesseract object after text has been recognized from image. Preventing memory leaks.

 

 

備忘：看過一篇博文提到：爲了提升效率須要對圖片進行預處理（二值化、灰度、傾斜校訂和圖片切割），傾斜校訂和圖片切割能夠用openCV的庫處理

博文連接：http://www.cocoachina.com/bbs/read.php?tid=123463 （該博文有demo這裏就很少鏈了）

 

比較有用的連接：

• FAQ：http://code.google.com/p/tesseract-ocr/wiki/FAQ
• 命令行程序幫助：http://tesseract-ocr.googlecode.com/svn/trunk/doc/tesseract.1.html
• 基於tesseract開發的軟件：http://code.google.com/p/tesseract-ocr/wiki/3rdParty
• tesseract提供的工具與各類語言的API：http://code.google.com/p/tesseract-ocr/wiki/AddOns