Research On PCB Netlist Extraction Based On CBCT Images

In recent years, a large number of advanced electronic equipments have been imported with non-technical, schematics of these devices are unrevealed. Maintenance of these equipments, especially the digestion and absorption of the design ideology of the equipments are very important to the development of electronic equipment with self intellectual property rights of our country. The schematic of equipments can be reverted by using PCB Reverse Engineering technology, which can meet the demand of maintenance and copy of these equipments. Howerver, current PCB Reverse Engineering technologies are contiguous, damaging or inefficient, to overcome these disadvantages, our PCB Reverse Engineering technology is achieved based on PCB 3D imaging using CBCT.The main research content of this paper is extracting the netlist from the PCB 3D image. Baesed on the characteristics of the PCB 3D image and combined with the demands of the practical applications, acquisition of the physical layer image of PCB, PCB elements recognition and PCB nelist extraction are studied in this paper. The main works of this thesis are as follows:1. A PCB physical layer image generation algorithm is presented. As PCB may distort to some extent during producing and welding, the physical layer image can not be obtained by directly slcing the PCB 3D image. Focusing on this problem, the proposed method found a surface approaching the physical layer based on the distribution of the track, and finally got the physical layer image. Experimental results show that the method can accurately extract the PCB two-dimensional images of the physical layers.2. Both SIFT detector and SURF descriptor are improved, and the image stitching has been implemented by these two improved algorithms respectively. Subject to the scan view of the CBCT, large-size PCB can only be scaned divisionally, then the image stitching algorithm is needed to get the entire physcial layer images. The detection time of the SIFT's detector is reduced by simplified the scale-space and interesting point selection. By improving the SURF descriptor, the matching speed between points is accelerated. The experiments show both algorithms achieve a satisfactory stitiching results, and SIFT has the advantage of precision while SURF has the advantage of speed.3. PCB elements extraction methods and PCB netlist generation algorithm are proposed. Pad, via and track, three main elements of PCB, are extracted by the combination of automated way and human-computer interaction. Based on the connnectivity expressed by the PCB element image, the subnets of the physical image can be obtained by labeling the connected area of the physical image. The subnets between physical layers can be merged by the characteristics that the vias connect with the physical layers. Experiments show that the netlist can be extracted from the PCB element image quickly and accurately.4. PCB netlist generation software based on PCB physical image is designed and implemented. The software includes two major modules: PCB Elements Extraction module and Netlist Generation module. This software has high performance of human-computer interaction and is easy to extend.