Procedures ( Computer Vision )
Our next concentration is to identify traces connection on the circuit board in order to determine paths. The system needs to find the connection from one place to another and determine what components are located along the path. The Computer Vision then combines the result from finding traces with Object Detection to regenerate the schematic design. It is noticeable that it has interconnecting inputs and outputs as it is the center of exchanging data and final result production.
To achieve that, the team needs to filter out all unnecessary components on the circuit to only identify traces. One of the techniques the team used is RGB Thresholding. Since the traces have a specific RGB value, it is efficient to set the RGB Threshold to that value in order to find traces and neglect the rest values. We used Contour Threshold Algorithm from the OpenCV framework onto the sample PCB image. The sample image is shown below.
After applying the Contour function onto this sample image, the team received the binary RGB value result as shown here.
It is observable that the traces are more easily identified and the rest of components are neglected. Even though on some traces there are minor distortions as seen in the lower right corner, for instance, they do not potentially affect the segmentation process, which will be applied afterwards.
The next part of the project is to apply object segmentation on the PCB Traces to capture the connection and ultimately combine the location of circuit elements from the results of object detection. Then, the team will investigate a solution of creating the symbolic appearance of each circuit component and placing it to the PCB trace. More progress will be reported once the next process is complete.