使用Tesseract OCR识别数字-优快云博客

之前一直想做一个人脸识别的项目，结果最后被调离那个项目组，最后不了了之。最近在家休息，一哥们就让帮忙采集一些数据，因为有效的数据被图片化了，就需要一些简单的数字识别，然后决定用tesseract-ocr来实现。

tesseract-ocr是一个OCR引擎，在1985年到1995年由HP实验室开发，现在在Google。tesseract-ocr 3.0发布，支持中文。开源协议Apache License 2.0，在商业项目中还是可以值得一用的。

OK.废话少说。代码贴上了。简单的测试了下，生产环境下自己封装下。OCR这些对资源的消耗特别严重，并且在现实情况下图片有噪点扭曲，失败的概率有点大,所以穷举的时候要注意对象代的管理。不过京东没有扭曲什么的，识别就简单多了。

 
    001using  System; 
 
    002using  System.Collections.Generic; 
 
    003using  System.Text; 
 
    004using  System.Drawing; 
 
    005using  System.IO; 
 
    006using  System.Linq; 
 
    007using  System.Net; 
 
    008using  System.Threading; 
 
    009  
 
    010using  tessnet2; 
 
    011namespace  OCR 
 
    012{ 
 
    013    class Program 
 
    014    {  
 
    015  
 
    016        static void Main(string[] args) 
 
    017        { 
 
    018            var imgpath = "http://image.58.com/showphone.aspx?t=v55&v=D438D3A47DA8384485F9AB46324E7BDB0"; 
 
    019            var ocr = new Tesseract(); 
 
    020            var result = new List<Word>(13); 
 
    021            var sb = new StringBuilder(20); 
 
    022            var ocrlength = 0; 
 
    023            var ocrcount = 0; 
 
    024            ocr.SetVariable("tessedit_char_whitelist", "￥-0123456789."); 
 
    025            ocr.Init(Path.Combine(Directory.GetCurrentDirectory(), @"Language"), "eng",  false); 
 
    026  
 
    027            /*识别58并验证是否电话号码*/
 
    028            Spot.Test(imgpath, ocr, out result); 
 
    029            ocrlength = result.Single<Word>().Text.Length; 
 
    030            while (ocrlength != 11) /// 检查是否匹配,实际环境中用正则 
 
    031            { 
 
    032                ocrcount++; 
 
    033                if (ocrcount % 5 != 0) 
 
    034                { 
 
    035                    Spot.Test(imgpath, ocr, out result); 
 
    036                    ocrlength = result.Single<Word>().Text.Length; 
 
    037                    while (ocrlength == 11) 
 
    038                    { 
 
    039                        Console.WriteLine("识别成功"); 
 
    040                        result.Each<Word>(r => Console.WriteLine(r.Text)); 
 
    041                        Console.ReadLine(); 
 
    042                        return; 
 
    043                    } 
 
    044                } 
 
    045                else
 
    046                { 
 
    047                    Console.WriteLine("为了缓解服务器的压力,稍微休息5秒之后继续识别"); 
 
    048                    Thread.Sleep(5000);//任务休息5秒 
 
    049                    GC.Collect(); 
 
    050                    GC.Collect(2); 
 
    051                } 
 
    052                sb.Remove(0, sb.Length); 
 
    053                result.Each<Word>(r => sb.Append(r.Text)); 
 
    054                Console.WriteLine("第{0}次错误识别结果是{1}", ocrcount, sb.ToString()); 
 
    055            } 
 
    056  
 
    057  
 
    058            /* 识别京东价格 
 
    059            Spot.Test(imgpath,ocr,out result); 
 
    060            result.Each<Word>(r => sb.Append(r.Text)); 
 
    061            Console.WriteLine(sb.Replace('S', '5').Replace('s', '5')); 
 
    062            Console.ReadLine(); 
 
    063             */
 
    064        } 
 
    065        } 
 
    066  
 
    067  
 
    068      
 
    069    public static class Spot 
 
    070    {  
 
    071        /// <summary> 
 
    072        /// 测试方法，实际环境中注意对象的回收。 
 
    073        /// </summary> 
 
    074        /// <remarks> 
 
    075        /// Spot.Test("","",""); 
 
    076        /// </remarks> 
 
    077        /// <param name="url">图片Url</param> 
 
    078        /// <param name="ocr">ocr的全局对象</param> 
 
    079        /// <param name="result">返回识别的结果</param> 
 
    080        public static void Test(string url,Tesseract ocr,out List<Word> result) 
 
    081        { 
 
    082            using (var wc = new WebClient()) 
 
    083            { 
 
    084                var img = wc.DownloadData(url); 
 
    085                using (var bitimage = new Bitmap(new  MemoryStream(img), true)) 
 
    086                { 
 
    087                    result = ocr.DoOCR(bitimage, Rectangle.Empty); 
 
    088                } 
 
    089            } 
 
    090        } 
 
    091    }  
 
    092  
 
    093      
 
    094    /// <summary> 
 
    095    ///  
 
    096    /// </summary> 
 
    097    public static class Linq 
 
    098    {  
 
    099        /// <summary> 
 
    100        /// 扩展 
 
    101        /// </summary> 
 
    102        /// <typeparam name="T"></typeparam> 
 
    103        /// <param name="col"></param> 
 
    104        /// <param name="handler"></param> 
 
    105        public static void Each<T>(this IEnumerable<T> col, Action<T> handler) 
 
    106        { 
 
    107            foreach (var item in col) 
 
    108                handler(item); 
 
    109        } 
 
    110    }  
 
    111  
 
    112}