Modified algorithm of unimodal thresholding for FPGA implementation

Thresholding is one of the basic digital image processing procedures that is commonly used for segmentation or edge detection. The final result of thresholding is significantly affected by the used threshold value. The several methods for determining the optimal threshold have been developed. Otsu's and Rosin's methods are the most used algorithms for computing the threshold value in non-real time image processing on a PC. These methods have not been used in FPGA circuits so far because of complicated calculation algorithm. In this paper we propose modification of Rosin's algorithm based on the geometric transformation of the histogram. Our approach simplifies calculations and allows FPGA implementation of Rosin's method. The simulations with three different images confirmed, that the proposed algorithm gives the same results as the original.