2D Target Object Recognition Tool
This section introduces the main functions and typical application scenarios of the 2D Target Object Recognition Tool.
Function Introduction
The 2D Target Object Recognition Tool is a visual debugging tool that integrates common vision recognition functions. It supports four major application scenarios: position-and-pick, placement correction, error-proofing inspection, and information reading.
-
Position and Pick: Uses a 2D camera to identify poses of same-type workpieces on the XOY plane (the Z-direction height must be consistent). It is suitable for picking non-stacked workpieces in trays and flat workpieces on conveyors.
-
Placement Correction: Before placing a workpiece, uses a 2D camera to detect its current position and then performs correction for accurate placement. It is suitable for scenarios with high placement-accuracy requirements, such as assembly.
-
Error-Proofing Inspection: Uses a 2D camera to identify workpiece status and verify whether it meets expectations, preventing missing, reversed, or incorrect assembly. It is suitable for front/back classification, presence/absence detection, deformation classification, misalignment/tilt detection, and quantity counting.
-
Information Reading: Integrates OCR and 1D/2D code recognition to read 1D codes, 2D codes, or character information on workpieces through a 2D camera, supplementing key identification data such as product model. It is suitable for material traceability and model identification.
Reading Guide
Configuration workflows for 2D target object recognition vary by scenario. For specific operations, read the following sections.
Prerequisites:
-
2D camera calibration is completed. If image distortion exists, perform distortion calibration first (applicable to all scenarios). For position-and-pick and placement-correction scenarios, extrinsic calibration must also be completed.
-
The camera can capture images normally, and image quality meets requirements. If camera parameters need adjustment, refer to 2D Camera Management to optimize parameters of the currently connected camera.
Position and Pick
In position-and-pick scenarios, target object recognition can be completed by either 2D Template Matching or 2D Blob Analysis.
-
2D Template Matching: If you need to search for and locate workpiece features that match templates in 2D images, choose 2D Template Matching. For details, refer to 2D Template Matching.
-
2D Blob Analysis: If you need to detect blobs in images and filter targets by geometric features, choose 2D Blob Analysis. For details, refer to 2D Blob Analysis.
Placement Correction
In placement-correction scenarios, you can recognize target objects for both fixed and non-fixed placement positions.
-
Fixed Placement Position: If the position and orientation of the placement position remain unchanged during runtime, choose recognition for fixed placement positions. For details, refer to Fixed Placement Position.
-
Non-Fixed Placement Position: If the actual position or orientation of the placement position varies in each run, choose recognition for non-fixed placement positions. For details, refer to Non-Fixed Placement Position.
| Before starting the detailed configuration workflow, for non-fixed placement scenarios, first configure a placement-position recognition project. The recognition method can be flexibly built based on actual needs. After recognition is completed, save the recognized placement pose to global variables correctly. For global variable configuration, refer to Global Variables. |
Error-Proofing Inspection
In error-proofing inspection scenarios, target object recognition can be completed through four methods: binary classification (front/back or present/absent), deformation classification, misalignment/tilt classification, and 2D template-matching counting.
-
Front/Back or Presence/Absence Classification: If you need to determine workpiece orientation (front/back) or presence/absence, choose binary classification. For details, refer to Front/Back or Presence/Absence Classification.
-
Deformation Classification: If you need to detect whether workpieces are deformed, choose deformation classification. For details, refer to Deformation Classification.
-
Misalignment/Tilt Classification: If you need to detect whether workpiece placement position or orientation is abnormal, choose misalignment/tilt classification. For details, refer to Misalignment/Tilt Classification.
-
2D Template-Matching Counting: If you need to recognize workpieces based on 2D template matching and count workpieces, choose 2D template-matching counting. For details, refer to 2D Template-Matching Counting.
Information Reading
In information-reading scenarios, target recognition can be completed through either 1D/2D code recognition or character recognition.
-
1D/2D Code Recognition: If you need to read 1D or 2D code information on targets, choose 1D/2D code recognition. For details, refer to 1D/2D Code Recognition.
-
Character Recognition: If you need to recognize printed, engraved, or inkjet characters on target surfaces, choose character recognition. For details, refer to Character Recognition.