Research and Implementation of FPGA-based Local Adaptive Thresholding

In view of the problems of the current fixed threshold algorithm, which is prone to clustering effect caused by uneven distribution of ambient light, unable to distinguish the foreground and background of the image, and unable to extract a large number of feature points, this paper adopts the innovative design of accelerating algorithm processing of verilog HDL hardware circuit platform. The innovation of the algorithm lies in using the average value of local extreme points and the texture feature of the whole image to adapt to the changes of complex environment in real time. The experimental results show that after the image is processed by the local adaptive image algorithm, the feature point information of the target contour can still be well extracted when the image is interfered by the uneven distribution of light. The similarity fitting of the feature point of the contour can reach 86%, which is improved compared with the fixed threshold of 73%. There are no obvious breakpoints, voids, white noise and other problems, and a relatively complete image is displayed. At the same time, the image target can be easily recognized when the background and foreground colors are switched.