<p>In this article, a new optimal multi-level thresholding algorithm for histogram-based image segmentation present. The proposed algorithm is an improved type of particle optimization algorithm. In this paper, the selection of image Thresholding done using a new method called Buzzard Optimization Algorithm (BUZO). Each particle (Buzzard) has two vectors in the search space, one is the ability vector, and the other is the position vector. The new position depends on these two vectors. The proposed BUZO algorithm has also solve these weaknesses of being stuck in local optimal points and early convergence and compensates for the failings of other algorithms. PSNR compares the proposed BUZO algorithm with PSO and MAFPSO algorithms as follows: In the image of the tree for the PSNR test of the BUZO algorithm in 5-level thresholding, it is 29.8, which is more different from the PSNR of the PSO and MAFPSO algorithms, which are 25 and 28, respectively. In addition, in another pirate test image, the PSNR of the BUZO algorithm in 5-level thresholding is 28.3, compared to the PSNR of PSO and MAFPSO algorithms, which are 25 and 26, respectively. In addition, in the pirate test image, the PSNR value of the BUZO algorithm in 2-level thresholding is 22.9, which compared to the PSNR of PSO algorithm 18 and MAFPSO algorithm 22.3. These results show the superiority of this algorithm with other methods.</p>