枝切法是一种高效的抗噪声相位展开方法,而最短枝切长度能够保证最优的相位展开结果.最短枝切长度问题属于组合优化问题,提出一种求解该问题的学习算法,将最短枝切长度问题的解视为个体,该算法通过个体之间的学习以及个体自身的变异实现进化,作用类似于遗传算法中的交叉算子以及变异算子.通过对多幅含噪声包裹相位图进行实验验证,该算法比传统的求解最短枝切长度问题的算法更快更优. Branch cut method is an efficient anti-noise phase unwrapping method, and the shortest branch length can ensure the optimal phase unwrapping results.The shortest branch length is a combinatorial optimization problem, a learning algorithm to solve the problem, the shortest The branch length problem is interpreted as an individual, which is evolved through learning between individuals and individual variation, which is similar to the crossover operator and mutation operator in genetic algorithm. Experiments show that this algorithm is faster and better than the traditional algorithm for solving the shortest branch length problem.