文档首页/ 文字识别 OCR/ API参考/ API/ 手写文字识别
更新时间:2024-06-19 GMT+08:00
分享

手写文字识别

功能介绍

识别文档中的手写文字、印刷文字信息,并将识别的结构化结果以JSON格式返回给用户。该接口的使用限制请参见约束与限制,详细使用指导请参见OCR服务使用简介章节。

图1 手写文字示例图

约束与限制

  • 只支持识别PNG、JPG、JPEG、BMP、TIFF格式图片。
  • 图像各边的像素大小在15px到8192px之间。
  • 图像中识别区域有效占比超过80%,保证所有文字及其边缘包含在图像内。
  • 支持图像任意角度的水平旋转(需开启方向检测)。
  • 目前不支持复杂背景(如户外自然场景、防伪水印等)和表格线扭曲图像的文字识别。
  • 文字书写越工整,识别率越高。

调用方法

请参见如何调用API

前提条件

在使用之前,需要您完成服务申请和认证鉴权,具体操作流程请参见开通服务认证鉴权章节。

用户首次使用需要先申请开通。服务只需要开通一次即可,后面使用时无需再次申请。如未开通服务,调用服务时会提示ModelArts.4204报错,请在调用服务前先进入控制台开通服务,并注意开通服务区域与调用服务的区域保持一致。

URI

POST /v2/{project_id}/ocr/handwriting

表1 路径参数

参数

是否必选

说明

endpoint

终端节点,即调用API的请求地址。

不同服务不同区域的endpoint不同,您可以从终端节点中获取。

例如,OCR服务在“华北-北京四”区域的“endpoint”“ocr.cn-north-4.myhuaweicloud.com”

project_id

项目ID,您可以从获取项目ID中获取。

请求参数

表2 请求Header参数

参数

是否必选

参数类型

描述

X-Auth-Token

String

用户Token。

用于获取操作API的权限。获取Token接口响应消息头中X-Subject-Token的值即为Token。

Content-Type

String

发送的实体的MIME类型,参数值为“application/json”。

Enterprise-Project-Id

String

企业项目ID。OCR支持通过企业项目管理(EPS)对不同用户组和用户的资源使用,进行分账。

获取方法:进入“企业项目管理”页面,单击企业项目名称,在企业项目详情页获取Enterprise-Project-Id(企业项目ID)。

企业项目创建步骤请参见用户指南。

说明:

创建企业项目后,在传参时,有以下三类场景。

  • 携带正确的ID,正常使用OCR服务,账单的企业项目会被分类到企业ID对应的企业项目中。
  • 携带格式正确但不存在的ID,正常使用OCR服务,账单的企业项目会显示对应不存在的企业项目ID。
  • 不携带ID或格式错误ID(包含特殊字符等),正常使用OCR服务,账单的企业项目会被分类到"default"中。
表3 请求Body参数

参数

是否必选

参数类型

说明

image

String

该参数与url二选一。

图片的Base64编码,要求Base64编码后大小不超过10MB。

图片最短边不小于15px,最长边不超过8192px,支持JPEG、JPG、PNG、BMP、TIFF格式。

图片Base64编码示例如/9j/4AAQSkZJRgABAg...,带有多余前缀会产生The image format is not supported报错。

url

String

该参数与image二选一。图片的url路径,目前支持:

  • 公网http/https url例如https://support.huaweicloud.com/api-ocr/zh-cn_image_0288038984.png
  • OBS提供的url,使用OBS数据需要进行授权。包括对服务授权、临时授权、匿名公开授权,详情参见配置OBS访问权限
说明:
  • 接口响应时间依赖于图片的下载时间,如果图片下载时间过长,会返回接口调用失败。
  • 请保证被检测图片所在的存储服务稳定可靠,推荐使用OBS服务存储图片数据。
  • url中不能存在中文字符,若存在,中文需要进行utf8编码。

quick_mode

Boolean

快速模式开关,针对单行文字图片(要求图片只包含一行文字,且文字区域占比超过50%),打开时可以更快返回识别内容。可选值如下所示。

  • true:打开快速模式
  • false:关闭快速模式

未传入该参数时默认为false,即关闭快速模式。

char_set

String

字符集设置,用户可以根据实际需要限定输出字符集范围。可选值如下所示。

  • digit:数字模式
  • letter:大小写字母模式
  • digit_letter:数字+字母模式
  • general:数字+字母+中文模式

未传入该参数时,默认为“general”模式。

detect_direction

Boolean

是否校正图片的倾斜角度,可选值如下。

  • true:校正图片的倾斜角度
  • false:不校正图片的倾斜角度

支持任意角度的校正,未传入该参数时默认为“false”

待识别图片如果存在倾斜,建议将此参数设置为“true”。

响应参数

根据识别的结果,可能有不同的HTTP响应状态码(status code)。例如,200表示API调用成功,400表示调用失败,详细的状态码和响应参数说明如下。

状态码: 200

表4 响应Body参数

参数

参数类型

描述

result

HandwritingResult object

识别结果。

调用失败时不返回此字段。

表5 HandwritingResult

参数

参数类型

描述

words_block_count

Integer

检测到的文字块数目。

words_block_list

Array of HandwritingItemsResponse objects

识别文字块列表。输出顺序从左到右,从上到下。

表6 HandwritingItemsResponse

参数

参数类型

描述

words

String

文字块识别结果。

type

String

说明该识别结果所属类型,返回值为“text”。

confidence

Float

字段的置信度,取值范围0~1。

置信度越大,本次识别的字段的可靠性越高,在统计意义上,置信度越大,准确率越高。

置信度由算法给出,不直接等价于字段的准确率。

location

Array<Array<Integer>>

文字块“words”的区域位置信息,列表形式,分别表示文字块顶点的(x,y)坐标;采用图像坐标系,坐标原点为图片左上角,x轴沿水平方向,y轴沿竖直方向。

状态码: 400

表7 响应Body参数

参数

参数类型

说明

error_code

String

调用失败时的错误码,具体请参见错误码

调用成功时不返回此字段。

error_msg

String

调用失败时返回的错误信息。

调用成功时不返回此字段。

请求示例

  • “endpoint”即调用API的请求地址,不同服务不同区域的“endpoint”不同,具体请参见终端节点

    例如,手写文字识别服务部署在“华北-北京四”区域的“endpoint”“ocr.cn-north-4.myhuaweicloud.com”“ocr.cn-north-4.myhuaweicloud.cn”,请求URL为“https://ocr.cn-north-4.myhuaweicloud.com/v2/{project_id}/ocr/handwriting”“project_id”为项目ID,获取方法请参见获取项目ID

  • 如何获取Token请参见认证鉴权
  • 传入手写文字图片的base64编码进行内容识别,识别范围包括数字、字母、中文,识别过程关闭快速模式并不校验图片倾斜角度
    POST https://{endpoint}/v2/{project_id}/ocr/handwriting 
     Request Header:   
     Content-Type: application/json   
     X-Auth-Token: MIINRwYJKoZIhvcNAQcCoIINODCCDTQCAQExDTALBglghkgBZQMEAgEwgguVBgkqhkiG...     
     Request Body:
     { 
        "image": "/9j/4AAQSkZJRgABAgEASABIAAD/4RFZRXhpZgAATU0AKgAAAA...",
        "quick_mode": false,
        "char_set": "general",
        "detect_direction": false
      }
  • 传入手写文字图片的url进行内容识别,识别范围包括数字、字母、中文,识别过程关闭快速模式并不校验图片倾斜角度
    POST https://{endpoint}/v2/{project_id}/ocr/handwriting 
     Request Header:   
     Content-Type: application/json   
     X-Auth-Token: MIINRwYJKoZIhvcNAQcCoIINODCCDTQCAQExDTALBglghkgBZQMEAgEwgguVBgkqhkiG...      
     Request Body:
     {
         "url":"https://BucketName.obs.xxxx.com/ObjectName",
         "quick_mode":false,
         "char_set": "general",
         "detect_direction": false
      }

响应示例

状态码:200

成功响应示例

{
    "result": {
        "words_block_count": 2,
        "words_block_list": [
            {
                "words": "大江东去",
                "type": "text",
                "confidence": 0.98,
                "location": [
                    [282, 45],
                    [461, 47],
                    [460, 77],
                    [280, 76]
                ]
            },
            {
                "words": "浪淘尽",
                "type": "text",
                "confidence": 0.99,
                "location": [
                    [949, 52],
                    [1095, 53],
                    [1100, 87],
                    [953, 86]
                ]
            }
        ]

    }
}

状态码:400

失败响应示例

{
    "error_code": "AIS.0103",
    "error_msg": "The image size does not meet the requirements."
}

SDK代码示例

SDK代码示例如下。

使用SDK前建议将SDK更新至最新版,防止本地旧版SDK无法使用最新的OCR功能。

  • 传入手写文字图片的base64编码进行内容识别,识别范围包括数字、字母、中文,识别过程关闭快速模式并不校验图片倾斜角度。
     1
     2
     3
     4
     5
     6
     7
     8
     9
    10
    11
    12
    13
    14
    15
    16
    17
    18
    19
    20
    21
    22
    23
    24
    25
    26
    27
    28
    29
    30
    31
    32
    33
    34
    35
    36
    37
    38
    39
    40
    41
    42
    43
    44
    45
    46
    47
    48
    49
    50
    51
    package com.huaweicloud.sdk.test;
    
    import com.huaweicloud.sdk.core.auth.ICredential;
    import com.huaweicloud.sdk.core.auth.BasicCredentials;
    import com.huaweicloud.sdk.core.exception.ConnectionException;
    import com.huaweicloud.sdk.core.exception.RequestTimeoutException;
    import com.huaweicloud.sdk.core.exception.ServiceResponseException;
    import com.huaweicloud.sdk.ocr.v1.region.OcrRegion;
    import com.huaweicloud.sdk.ocr.v1.*;
    import com.huaweicloud.sdk.ocr.v1.model.*;
    
    
    public class RecognizeHandwritingSolution {
    
        public static void main(String[] args) {
            // The AK and SK used for authentication are hard-coded or stored in plaintext, which has great security risks. It is recommended that the AK and SK be stored in ciphertext in configuration files or environment variables and decrypted during use to ensure security.
            // In this example, AK and SK are stored in environment variables for authentication. Before running this example, set environment variables CLOUD_SDK_AK and CLOUD_SDK_SK in the local environment
            String ak = System.getenv("CLOUD_SDK_AK");
            String sk = System.getenv("CLOUD_SDK_SK");
    
            ICredential auth = new BasicCredentials()
                    .withAk(ak)
                    .withSk(sk);
    
            OcrClient client = OcrClient.newBuilder()
                    .withCredential(auth)
                    .withRegion(OcrRegion.valueOf("<YOUR REGION>"))
                    .build();
            RecognizeHandwritingRequest request = new RecognizeHandwritingRequest();
            HandwritingRequestBody body = new HandwritingRequestBody();
            body.withDetectDirection(false);
            body.withCharSet("general");
            body.withQuickMode(false);
            body.withImage("/9j/4AAQSkZJRgABAgEASABIAAD/4RFZRXhpZgAATU0AKgAAAA...");
            request.withBody(body);
            try {
                RecognizeHandwritingResponse response = client.recognizeHandwriting(request);
                System.out.println(response.toString());
            } catch (ConnectionException e) {
                e.printStackTrace();
            } catch (RequestTimeoutException e) {
                e.printStackTrace();
            } catch (ServiceResponseException e) {
                e.printStackTrace();
                System.out.println(e.getHttpStatusCode());
                System.out.println(e.getRequestId());
                System.out.println(e.getErrorCode());
                System.out.println(e.getErrorMsg());
            }
        }
    }
    
  • 传入手写文字图片的url进行内容识别,识别范围包括数字、字母、中文,识别过程关闭快速模式并不校验图片倾斜角度
     1
     2
     3
     4
     5
     6
     7
     8
     9
    10
    11
    12
    13
    14
    15
    16
    17
    18
    19
    20
    21
    22
    23
    24
    25
    26
    27
    28
    29
    30
    31
    32
    33
    34
    35
    36
    37
    38
    39
    40
    41
    42
    43
    44
    45
    46
    47
    48
    49
    50
    51
    package com.huaweicloud.sdk.test;
    
    import com.huaweicloud.sdk.core.auth.ICredential;
    import com.huaweicloud.sdk.core.auth.BasicCredentials;
    import com.huaweicloud.sdk.core.exception.ConnectionException;
    import com.huaweicloud.sdk.core.exception.RequestTimeoutException;
    import com.huaweicloud.sdk.core.exception.ServiceResponseException;
    import com.huaweicloud.sdk.ocr.v1.region.OcrRegion;
    import com.huaweicloud.sdk.ocr.v1.*;
    import com.huaweicloud.sdk.ocr.v1.model.*;
    
    
    public class RecognizeHandwritingSolution {
    
        public static void main(String[] args) {
            // The AK and SK used for authentication are hard-coded or stored in plaintext, which has great security risks. It is recommended that the AK and SK be stored in ciphertext in configuration files or environment variables and decrypted during use to ensure security.
            // In this example, AK and SK are stored in environment variables for authentication. Before running this example, set environment variables CLOUD_SDK_AK and CLOUD_SDK_SK in the local environment
            String ak = System.getenv("CLOUD_SDK_AK");
            String sk = System.getenv("CLOUD_SDK_SK");
    
            ICredential auth = new BasicCredentials()
                    .withAk(ak)
                    .withSk(sk);
    
            OcrClient client = OcrClient.newBuilder()
                    .withCredential(auth)
                    .withRegion(OcrRegion.valueOf("<YOUR REGION>"))
                    .build();
            RecognizeHandwritingRequest request = new RecognizeHandwritingRequest();
            HandwritingRequestBody body = new HandwritingRequestBody();
            body.withDetectDirection(false);
            body.withCharSet("general");
            body.withQuickMode(false);
            body.withUrl("https://BucketName.obs.myhuaweicloud.com/ObjectName");
            request.withBody(body);
            try {
                RecognizeHandwritingResponse response = client.recognizeHandwriting(request);
                System.out.println(response.toString());
            } catch (ConnectionException e) {
                e.printStackTrace();
            } catch (RequestTimeoutException e) {
                e.printStackTrace();
            } catch (ServiceResponseException e) {
                e.printStackTrace();
                System.out.println(e.getHttpStatusCode());
                System.out.println(e.getRequestId());
                System.out.println(e.getErrorCode());
                System.out.println(e.getErrorMsg());
            }
        }
    }
    
  • 传入手写文字图片的base64编码进行内容识别,识别范围包括数字、字母、中文,识别过程关闭快速模式并不校验图片倾斜角度。
     1
     2
     3
     4
     5
     6
     7
     8
     9
    10
    11
    12
    13
    14
    15
    16
    17
    18
    19
    20
    21
    22
    23
    24
    25
    26
    27
    28
    29
    30
    31
    32
    33
    34
    35
    # coding: utf-8
    
    from huaweicloudsdkcore.auth.credentials import BasicCredentials
    from huaweicloudsdkocr.v1.region.ocr_region import OcrRegion
    from huaweicloudsdkcore.exceptions import exceptions
    from huaweicloudsdkocr.v1 import *
    
    if __name__ == "__main__":
        # The AK and SK used for authentication are hard-coded or stored in plaintext, which has great security risks. It is recommended that the AK and SK be stored in ciphertext in configuration files or environment variables and decrypted during use to ensure security.
        # In this example, AK and SK are stored in environment variables for authentication. Before running this example, set environment variables CLOUD_SDK_AK and CLOUD_SDK_SK in the local environment
        ak = os.getenv("CLOUD_SDK_AK")
        sk = os.getenv("CLOUD_SDK_SK")
    
        credentials = BasicCredentials(ak, sk) \
    
        client = OcrClient.new_builder() \
            .with_credentials(credentials) \
            .with_region(OcrRegion.value_of("<YOUR REGION>")) \
            .build()
    
        try:
            request = RecognizeHandwritingRequest()
            request.body = HandwritingRequestBody(
                detect_direction=False,
                char_set="general",
                quick_mode=False,
                image="/9j/4AAQSkZJRgABAgEASABIAAD/4RFZRXhpZgAATU0AKgAAAA..."
            )
            response = client.recognize_handwriting(request)
            print(response)
        except exceptions.ClientRequestException as e:
            print(e.status_code)
            print(e.request_id)
            print(e.error_code)
            print(e.error_msg)
    
  • 传入手写文字图片的url进行内容识别,识别范围包括数字、字母、中文,识别过程关闭快速模式并不校验图片倾斜角度
     1
     2
     3
     4
     5
     6
     7
     8
     9
    10
    11
    12
    13
    14
    15
    16
    17
    18
    19
    20
    21
    22
    23
    24
    25
    26
    27
    28
    29
    30
    31
    32
    33
    34
    35
    # coding: utf-8
    
    from huaweicloudsdkcore.auth.credentials import BasicCredentials
    from huaweicloudsdkocr.v1.region.ocr_region import OcrRegion
    from huaweicloudsdkcore.exceptions import exceptions
    from huaweicloudsdkocr.v1 import *
    
    if __name__ == "__main__":
        # The AK and SK used for authentication are hard-coded or stored in plaintext, which has great security risks. It is recommended that the AK and SK be stored in ciphertext in configuration files or environment variables and decrypted during use to ensure security.
        # In this example, AK and SK are stored in environment variables for authentication. Before running this example, set environment variables CLOUD_SDK_AK and CLOUD_SDK_SK in the local environment
        ak = os.getenv("CLOUD_SDK_AK")
        sk = os.getenv("CLOUD_SDK_SK")
    
        credentials = BasicCredentials(ak, sk) \
    
        client = OcrClient.new_builder() \
            .with_credentials(credentials) \
            .with_region(OcrRegion.value_of("<YOUR REGION>")) \
            .build()
    
        try:
            request = RecognizeHandwritingRequest()
            request.body = HandwritingRequestBody(
                detect_direction=False,
                char_set="general",
                quick_mode=False,
                url="https://BucketName.obs.myhuaweicloud.com/ObjectName"
            )
            response = client.recognize_handwriting(request)
            print(response)
        except exceptions.ClientRequestException as e:
            print(e.status_code)
            print(e.request_id)
            print(e.error_code)
            print(e.error_msg)
    
  • 传入手写文字图片的base64编码进行内容识别,识别范围包括数字、字母、中文,识别过程关闭快速模式并不校验图片倾斜角度。
     1
     2
     3
     4
     5
     6
     7
     8
     9
    10
    11
    12
    13
    14
    15
    16
    17
    18
    19
    20
    21
    22
    23
    24
    25
    26
    27
    28
    29
    30
    31
    32
    33
    34
    35
    36
    37
    38
    39
    40
    41
    42
    43
    44
    45
    package main
    
    import (
    	"fmt"
    	"github.com/huaweicloud/huaweicloud-sdk-go-v3/core/auth/basic"
        ocr "github.com/huaweicloud/huaweicloud-sdk-go-v3/services/ocr/v1"
    	"github.com/huaweicloud/huaweicloud-sdk-go-v3/services/ocr/v1/model"
        region "github.com/huaweicloud/huaweicloud-sdk-go-v3/services/ocr/v1/region"
    )
    
    func main() {
        // The AK and SK used for authentication are hard-coded or stored in plaintext, which has great security risks. It is recommended that the AK and SK be stored in ciphertext in configuration files or environment variables and decrypted during use to ensure security.
        // In this example, AK and SK are stored in environment variables for authentication. Before running this example, set environment variables CLOUD_SDK_AK and CLOUD_SDK_SK in the local environment
        ak := os.Getenv("CLOUD_SDK_AK")
        sk := os.Getenv("CLOUD_SDK_SK")
    
        auth := basic.NewCredentialsBuilder().
            WithAk(ak).
            WithSk(sk).
            Build()
    
        client := ocr.NewOcrClient(
            ocr.OcrClientBuilder().
                WithRegion(region.ValueOf("<YOUR REGION>")).
                WithCredential(auth).
                Build())
    
        request := &model.RecognizeHandwritingRequest{}
    	detectDirectionHandwritingRequestBody:= false
    	charSetHandwritingRequestBody:= "general"
    	quickModeHandwritingRequestBody:= false
    	imageHandwritingRequestBody:= "/9j/4AAQSkZJRgABAgEASABIAAD/4RFZRXhpZgAATU0AKgAAAA..."
    	request.Body = &model.HandwritingRequestBody{
    		DetectDirection: &detectDirectionHandwritingRequestBody,
    		CharSet: &charSetHandwritingRequestBody,
    		QuickMode: &quickModeHandwritingRequestBody,
    		Image: &imageHandwritingRequestBody,
    	}
    	response, err := client.RecognizeHandwriting(request)
    	if err == nil {
            fmt.Printf("%+v\n", response)
        } else {
            fmt.Println(err)
        }
    }
    
  • 传入手写文字图片的url进行内容识别,识别范围包括数字、字母、中文,识别过程关闭快速模式并不校验图片倾斜角度
     1
     2
     3
     4
     5
     6
     7
     8
     9
    10
    11
    12
    13
    14
    15
    16
    17
    18
    19
    20
    21
    22
    23
    24
    25
    26
    27
    28
    29
    30
    31
    32
    33
    34
    35
    36
    37
    38
    39
    40
    41
    42
    43
    44
    45
    package main
    
    import (
    	"fmt"
    	"github.com/huaweicloud/huaweicloud-sdk-go-v3/core/auth/basic"
        ocr "github.com/huaweicloud/huaweicloud-sdk-go-v3/services/ocr/v1"
    	"github.com/huaweicloud/huaweicloud-sdk-go-v3/services/ocr/v1/model"
        region "github.com/huaweicloud/huaweicloud-sdk-go-v3/services/ocr/v1/region"
    )
    
    func main() {
        // The AK and SK used for authentication are hard-coded or stored in plaintext, which has great security risks. It is recommended that the AK and SK be stored in ciphertext in configuration files or environment variables and decrypted during use to ensure security.
        // In this example, AK and SK are stored in environment variables for authentication. Before running this example, set environment variables CLOUD_SDK_AK and CLOUD_SDK_SK in the local environment
        ak := os.Getenv("CLOUD_SDK_AK")
        sk := os.Getenv("CLOUD_SDK_SK")
    
        auth := basic.NewCredentialsBuilder().
            WithAk(ak).
            WithSk(sk).
            Build()
    
        client := ocr.NewOcrClient(
            ocr.OcrClientBuilder().
                WithRegion(region.ValueOf("<YOUR REGION>")).
                WithCredential(auth).
                Build())
    
        request := &model.RecognizeHandwritingRequest{}
    	detectDirectionHandwritingRequestBody:= false
    	charSetHandwritingRequestBody:= "general"
    	quickModeHandwritingRequestBody:= false
    	urlHandwritingRequestBody:= "https://BucketName.obs.myhuaweicloud.com/ObjectName"
    	request.Body = &model.HandwritingRequestBody{
    		DetectDirection: &detectDirectionHandwritingRequestBody,
    		CharSet: &charSetHandwritingRequestBody,
    		QuickMode: &quickModeHandwritingRequestBody,
    		Url: &urlHandwritingRequestBody,
    	}
    	response, err := client.RecognizeHandwriting(request)
    	if err == nil {
            fmt.Printf("%+v\n", response)
        } else {
            fmt.Println(err)
        }
    }
    

更多编程语言的SDK代码示例,请参见API Explorer的代码示例页签,可生成自动对应的SDK代码示例。

状态码

状态码

描述

200

成功响应示例

400

失败响应示例

状态码请参见状态码

错误码

错误码请参见错误码

相关文档