Threshold segmentation of PCB defect image grid based on finite difference dispersion for providing accuracy in the IoT based data of smart cities

The image segmentation is a process of segregating image into manageable partitions or segments. Image segmentation involves segmentation of foreground from background, based on similarities in features, color or shape. Segmentation is the foremost step in the scrutiny of the printed circuit boards. The segmentation technique basically partitions the PCB into small chunks containing primeval PCB sub-patterns. The conventional threshold segmentation technique considers the effect of image noise and grayscale in the threshold segmentation operation of the PCB (printed circuit board). In this paper, we have tried to overcome the problem of conventional segmentation for PCB such as grayscale overlapping and noise related issues in the segmented images. The basic formula of the optimal threshold operation is given to the threshold segmentation algorithm, the triangular image grid model is designed, the finite difference discrete method is exploited, the optimal threshold calculation formula is devised, and an attempt is made to get the best threshold for segmentation of PCB. Through experimental analysis for the PCB circuit image, two-dimensional OSTU threshold segmentation method, two-dimensional Fisher threshold segmentation method and the proposed research method are used to perform image grid threshold segmentation and to evaluate the performance of the proposed and the existing methods. The experimental results show that the threshold segmentation values obtained by the finite difference based threshold method has low image-noise effect, the gray scale overlap is also minimized, and segmentation accuracy is more as compare to other benchmarked threshold methods.