Correcting recognition and document filling errors
The document editor allows you to check the recognition accuracy and correct wrongly recognized characters, as well as to correct and add data missed when filling the document.
In the simplest case, the process of correcting errors looks like this:
- Click to go to the next error.
- Check the field value and correct it if necessary.
- Click the Enter. The station will go to the next field that needs checking.
Note: You can use the Tab button to navigate the fields on the data form.
- Repeat steps 2-3 until all fields are corrected.
- If there is a field that was not matched, select its region. Depending on the settings specified in the Document Definition, the field value will be either added from cache or you will need to re-recognize the region on the server.
- Check the rules on the server clicking the Check Rules button. If rule errors are found, correct them.
Document form contains the list of fields available for editing and the data extracted from those fields. When filling out the document and recognizing its data, the following errors may occur:
- rule execution or check errors (fields with such errors are highlighted in red);
- rule warnings or non-matched fields (fields with such errors are highlighted in yellow);
- requires verification (fields with such errors have a red underline).
For some fields, the system automatically prompts possible values (e.g., for a summarizing result field or a key field which recurs on different pages). In the lower right corner of such fields you will see this icon: clicking on which opens the list of possible values suggest by FlexiCapture according to the configured rules.
The Check Rules button initializes a server-side rule check. Most of the rules are checked locally in the Web Verification Station immediately after a field value is modified. However, there is a range of exceptions that require you to click the Check Rules button after modifying field values:
- if value normalization is applied,
- if number and text values of a number are being compared,
- for script rules that use objects that are not supported for a local rule check on a Web Verification Station. To check whether a script will be executed without calling a rule check, make sure that all objects used in the script are supported. The necessary information can be found in the description of each object: if an object is not supported or is only partially supported, this is specified in the description. If a script object is not supported, an error for the rule will be displayed during the verification process. The error will suggest that you click Check Rules.
Note: By default, errors of all types are displayed in the errors window. To hide errors of a certain type (format, rules, matching or region), click its icon in the errors view.
Page image is located in the central pane of the task window. Image scale can be changed using the mouse wheel or using the Zoom tool:
- "+" - zooms in;
- "-" - zooms out;
- "↔" - fits the page to the width of the screen
- "↕" - fits the page to the height of the screen.
Hover the mouse pointer over the pane, hold down the Ctrl key, and scroll the wheel. Scrolling the mouse wheel without holding down the Ctrl key will to move the image vertically. By default, the image part with a field selected in the data area is displayed.
The page image contains field regions which indicate the area where data recognition is performed.
If for some fields the regions were not matched or were matched incorrectly, you may need to draw the field region or change its borders or location. It is not always necessary to edit the area - sometimes it can be faster to enter the value in the data form manually.
Note: The drawback of entering value manually without creating a region is that, for example, when you export document to PDF with a text layer, the field value will not be associated with the location of text on the image.
If you need to create or edit the field region, then depending on Document Definition settings two variants are possible:
- In Document Definition properties, caching results of full text recognition is enabled.
Note: Full text recognition is performed only before matching flexible Document Definitions. Semi-structured document are the documents whose structure and the sections order can differ in different documents.
If cashing is used, then when you create or edit the region, recognized value will be added to the field from cache. If you edit field value manually, further operation with the region will not lead to changing the field value in order to not lose changes the made changes.
- If caching is not used, then after creating or editing the region, it must be recognized on the server. To send new or edited regions to recognition, click the Check Rules button. After the rule check, new or edited regions will be re-recognized.
Working with tables
If a table region is selected on a document image, you can use the tools located in the top left corner of the document image pane to perform the following actions:
- To add a vertical or horizontal separator, do one of the following:
- place the mouse pointer inside the table region, press and hold down the left mouse button, and draw a separator.
- select either the (vertical separator) tool or (horizontal separator) tool, position the mouse pointer inside the table region where you want to add a separator and left-click.
- To delete a separator, select the tool, place the mouse pointer inside the table region and left-click the separator that you want to delete.
Errors area is located at the bottom right part of the page. It shows all the errors detected in the batch. For different error types, the following icons are used:
- - Document assembly error. If the error is associated with mismatch of the key fields the fields values under each page are displayed in red. If the error is associated with an uncertainly recognized page present in a document, the name of this page and the name of the parent document will be highlighted in red.
- / - single-field / multi-field rule error;
- / - single-field / multi-field rule warning, or the document has non-matched fields.
- - field requires verification;
- - Unmatched region. Such an error means that the option Should be matched is selected in the field properties.
If necessary, you can hide information about the errors of particular type clicking the corresponding button at the top of the error area. It can be useful if Operator wants to process error of specific type.