QpixControl2 — 集成先进AI功能，重新定义硬度测试 自动、准确地检测硬度测试印痕——即使是复杂表面
这种图像评估用于硬度测试的所有领域，通常能提高识别率，发现图像中的压痕，并提升评估和分析的质量与准确性。 基于人工智能的图像评估显著提高了硬度测试中压痕检测的质量。
QAI 能为粗糙、研磨和腐蚀表面提供更大的附加值。特别是在应对复杂材料表面或腐蚀表面时，识别率能够大幅提高。
QAI 图像评估已完全集成到QpixControl2 操作软件中，并取代了现有的图像识别算法。
使用QAI图像识别技术也改进了设备的重复性和系统偏差。评估的准确性对设备的相对重复性有着极大影响。
传统评估与QAI评估的比较
在701/HV1硬度块上打90个测试点。针对相同的90个压痕，采用不同的评估模式进行测量。
| 平均值 | 范围 |
| 700,04 | 24,90 |
| 最小硬度值 | 最大硬度值 |
| 688,80 | 713,70 |
| 标准差 | 结果正常 |
| 5,88 | 90 |
| 平均值 | 范围 |
| 701,50 | 16,40 |
| 最小硬度值 | 最大硬度值 |
| 692,50 | 708,90 |
| 标准差 | 结果正常 |
| 3,47 | 90 |
AI人工智能及其图像识别功能仅在本地电脑上运行，且仅在 QpixControl2 软件内运行，所有数据均为离线状态，无需互联网连接。 人工智能模型无法自行发展和学习；此功能和工作只能由 QATM 负责进行，以确保设备上仅使用经过认证的 QAI。硬度计作为精密检测设备，必须符合标准，因此相关结果必须由QATM进行验证。 所有测试数据均存储在本地 PC 和软件中，不会与 QATM 进行数据交换。QAI测试结果始终保持一致性。
100%离线解决方案
100%本地数据
机器上的 QAI 没有持续进行开发
NO. The AI-based image recognition does not affect the optical system. The magnification, camera, and lenses remain unchanged. QAI analyzes the captured image and detects the hardness test indentation. The evaluation and measurement process follow the same principles as conventional hardness testing software.
NO. The relevant standards (DIN EN ISO, ASTM) specify requirements for sample preparation but do not define surface quality parameters such as roughness values (Ra/Rz). In general, the surface should be prepared appropriately for the Vickers hardness test, depending on the applied load. The indentation and its edges must be clearly visible.
Possibly, yes. QAI image evaluation can detect hardness indentations even on lower-quality surfaces. We recommend maintaining your current preparation process initially. However, step-by-step optimization is possible and should be validated accordingly.
Important note: The customer is responsible for defining and verifying their process. QATM can provide guidance and support.
YES. Technically and from a software perspective, direct hardness testing on etched surfaces is possible. QAI image evaluation can achieve very good detection rates even in these cases. However, standards recommend performing hardness tests on non-etched surfaces. The final responsibility for process validation lies with the customer.
NO. The AI and image recognition operate entirely locally on the PC within the QpixControl2 software. All data remains offline, and no internet access is required.
NO. The AI model cannot develop and learn itself independently. In the case that the QAI software cannot recognize hardness test indentations, there is the possibility to relearn the QAI by QATM.
