MicroLED defect detection method based on unsupervised learning
A defect detection and model technology, applied in the field of defect detection, can solve the problems of unbalanced positive and negative samples, difficult to define and create, difficult to solve, etc., to achieve the effect of improving accuracy and rich semantic information
Active Publication Date: 2022-05-17
利晶微电子技术(江苏)有限公司
View PDF6 Cites 3 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
However, methods based on supervised learning have some disadvantages: one is that they require datasets with real labels for each chip, which requires hard work
In addition, in the actual industrial process, compared with normal chips, defective chips rarely appear, and the problem of positive and negative sample imbalance hinders the effective training of DNN models.
Although the class positive and negative sample imbalance problem can be solved by deliberately creating defects or using data augmentation methods, it is difficult to define and create all possible defect modes, which is difficult to solve in real industrial processes
Therefore, the above shortcomings make it difficult to implement the current visual inspection method based on supervised learning and the results are not accurate enough.
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0041] The specific embodiments of the present invention will be further described below in conjunction with the accompanying drawings.
[0042] This application discloses a MicroLED defect detection method based on unsupervised learning, please combine figure 1 As shown in the flow chart, the method includes the following two parts: the model training part and the model application part, which are respectively introduced as follows:
[0043] 1. The model training part is used for training to obtain the defect detection model.
[0044] Including the following steps, please combine figure 1 :
[0045] Step 102, acquiring a sample data set, which includes normal sample images of normal MicroLED sample chips and abnormal sample images of abnormal MicroLED sample chips with defects.
[0046]In actual implementation, combined with the fact that defective chips rarely appear in the industry, the normal sample images in the sample data set are generally much larger than the abnorm...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The invention discloses a MicroLED defect detection method based on unsupervised learning, and relates to the field of defect detection, and the method comprises the steps: carrying out the model pre-training of a residual error convolution auto-encoder model through a normal sample image and an abnormal sample image which are subjected to image preprocessing; the residual error convolution auto-encoder model comprises an encoder formed based on a residual error convolution module and a decoder formed based on a residual error transpose convolution module, mapping the output of the encoder of the pre-trained residual error convolution auto-encoder model to a potential space and fitting to a hyperspherical surface, and optimizing the potential space by using an objective function, according to the method, the defect detection model is obtained through training, the defect detection model with higher robustness can be obtained so as to realize automatic defect detection of the MicroLED chip, and the method does not need to mark data and has robustness for the class imbalance problem.
Description
technical field [0001] The invention relates to the field of defect detection, in particular to a MicroLED defect detection method based on unsupervised learning. Background technique [0002] Light-emitting diodes (LEDs) are widely used in various fields such as displays, vehicles, and medical equipment. Demand for LEDs is increasing due to their high efficiency, low power consumption, long lifespan, and environmental protection characteristics. However, manufacturing defects destroy these advantages and bring huge losses to manufacturers, such as time and cost, so to make up for this defect, more accurate and faster inspection of LEDs is required. [0003] At present, the common defect inspection methods of LED chips mainly include automatic optical inspection (AOI), photoluminescence (PL) inspection and electroluminescence (EL) inspection. Among them, AOI is a non-contact visual inspection that detects surface defects on wafers or chips and prevents damage caused by con...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More IPC IPC(8): G06T7/00G06N3/04G06N3/08G06T5/30
CPCG06T7/0006G06T5/30G06N3/08G06T2207/20024G06T2207/30148G06N3/045
Inventor周佳潘彤郭震撼曹晖袁廷翼王杨杨夏天鲍涛
Owner利晶微电子技术(江苏)有限公司