Need better scanning performance?

Get a FREE customized reading template.

CaptureVisionTemplate Object

A CaptureVisionTemplate object is the entry point of a parameter template in Dynamsoft Capture Vision (DCV) SDK.

Example

{
    "Name": "CV_0",
    "ImageSourceName": "ISA_0",
    "ImageROIProcessingNameArray": ["TA_0"],
    "SemanticProcessingNameArray": ["SP_0"],
    "OutputOriginalImage": 0,
    "MaxParallelTasks": 4,
    "MinImageCaptureInterval": 0,
    "Timeout": 500
}

Parameters

Input Source Configuration

The ImageSourceName parameter references an ImageSource object defining the image input source. When DCV starts capturing, it parses the ImageSource parameter, converts it into an Image Source Adapter (ISA) object, and continuously obtains images from it.

Captured Output Configuration

OutputOriginalImage, ImageROIProcessingNameArray, and SemanticProcessingNameArray control the captured output, organized as a CapturedResult interface in DCV. CapturedResult represents a set of all captured result items on an image. Each type of result item represents the output of different task types. The following figure lists the relationship between DCV output parameters and output results.

Parameter and result relationships:

• ImageROIProcessingNameArray produces BarcodeResultItem, TextLineResultItem, DetectedQuadResultItem, and NormalizedImageResultItem
• SemanticProcessingNameArray produces ParsedResultItem
• ImageROIProcessingNameArray references one or more TargetROIDef objects
• SemanticProcessingNameArray references one or more SemanticProcessing objects

Core Design of TargetROIDef Object

The TargetROIDef object specifies one or more recognition tasks to be performed on regions of interest (ROIs) within an image. In simple terms:

TargetROIDef = Recognition Task Definition + Spatial Location Definition

Key Concepts

Workflow Design Based on Reference/Target Regions

Using recursive definition of reference relationships between TargetROIDef objects, complex workflows can be constructed for complex scenarios.

Example: Read barcodes below P/N and above L/N in the image. Positional dependencies:

• Target barcode location (blue box) can be calculated from text line locations (P/N and L/N)
• Text lines (P/N and L/N) are on the first and fifth lines inside the label border (green box)

The following json is a parameter template fragment that configures ROI dependencies to solve the above problems.

{
    "TargetROIDefOptions": [
        {
            "Name": "ddn_roi",
            "TaskSettingNameArray": ["ddn_task"],
            "Location": null
        },
        {
            "Name": "dlr_roi",
            "TaskSettingNameArray": ["dlr_task"],
            "Location": {
                "ReferenceObjectFilter": {
                    "ReferenceTargetROIDefNameArray": ["ddn_roi"]
                },
                "Offset": null
            }
        },
        {
            "Name": "dbr_roi1",
            "TaskSettingNameArray": ["dbr_task"],
            "Location": {
                "ReferenceObjectFilter": {
                    "ReferenceTargetROIDefNameArray": ["dlr_roi"]
                },
                "Offset": {
                    "MeasuredByPercentage": 0,
                    "ReferenceYAxis": {"AxisType": "AT_EDGE", "EdgeIndex": 2},
                    "FirstPoint": [0, 10],
                    "SecondPoint": [100, 10],
                    "ThirdPoint": [100, 50],
                    "FourthPoint": [0, 50]
                }
            }
        },
        {
            "Name": "dbr_roi2",
            "TaskSettingNameArray": ["dbr_task"],
            "Location": {
                "ReferenceObjectFilter": {
                    "ReferenceTargetROIDefNameArray": ["dlr_roi"]
                },
                "Offset": {
                    "MeasuredByPercentage": 0,
                    "ReferenceYAxis": {"AxisType": "AT_EDGE", "EdgeIndex": 0},
                    "FirstPoint": [0, -50],
                    "SecondPoint": [100, -50],
                    "ThirdPoint": [100, -10],
                    "FourthPoint": [0, -10]
                }
            }
        }
    ]
}

In this configuration:

• Four TargetROIDef objects: ddn_roi, dlr_roi, dbr_roi1, and dbr_roi2
• ddn_roi depends on the entire image region
• dlr_roi depends on ddn_roi with null offset (same region)
• dbr_roi1 depends on dlr_roi with downward offset
• dbr_roi2 depends on dlr_roi with upward offset

Construct a Dependency Graph

When DCV parses TargetROIDef objects, it constructs a directed dependency graph based on configured dependencies. Tasks execute in the order specified by the dependency graph.

Filter Out the Desired Reference Regions

In practical applications, a TargetROIDef may want to depend only on reference regions from another TargetROIDef that meet specific filtering conditions, such as text matching a regular expression or barcodes meeting a specific format requirements.

Based on the previous example, regular expression filtering conditions can be added to dbr_roi1 and dbr_roi2:

{
    "TargetROIDefOptions": [
        {
            "Name": "roi_dbr1",
            "TaskSettingNameArray": ["dbr_task"],
            "Location": {
                "ReferenceObjectFilter": {
                    "ReferenceTargetROIDefNameArray": ["roi_dlr"],
                    "TextLineFilteringCondition": {
                        "LineStringRegExPattern": "^P/N"
                    }
                },
                "Offset": {}
            }
        },
        {
            "Name": "roi_dbr2",
            "TaskSettingNameArray": ["dbr_task"],
            "Location": {
                "ReferenceObjectFilter": {
                    "ReferenceTargetROIDefNameArray": ["roi_dlr"],
                    "TextLineFilteringCondition": {
                        "LineStringRegExPattern": "^L/N"
                    }
                },
                "Offset": {}
            }
        }
    ]
}

Filtering logic:

• roi_dbr1 depends only on regions from roi_dlr where text matches pattern ^P/N
• roi_dbr2 depends only on regions from roi_dlr where text matches pattern ^L/N
• Reference regions not meeting filter criteria are discarded

For more details about filtering reference objects, refer to ReferenceObjectFilter.

Core Design of SemanticProcessing Object

The SemanticProcessing object specifies tasks to analyze and extract information from image ROI processing results. The workflow involves the following concepts:

Prerequisites

A SemanticProcessing object takes effect when its name is referenced in CaptureVisionTemplate.SemanticProcessingNameArray.

Launch Timing

The semantic process triggers only after all recognition tasks referenced in CaptureVisionTemplate.ImageROIProcessingNameArray have been completed.

Data Filtering

Data filtering selects relevant sources, such as label text matching a specific regular expression or barcodes meeting a specific format. Use ReferenceObjectFilter to specify data filtering criteria.

Task Execution

This is the main part of the workflow where actual tasks are defined. Use TaskSettingNameArray to specify tasks by referencing a CodeParserTaskSetting object name.

Results Reporting

Currently, semantic-processing supports code parsing tasks. Results are returned with the OnParsedResultsReceived callback.