An improved algorithm based on oppositional searching and chaos mutation is proposed in order to deal with the deficiencies of the traditional krill herd optimization algorithm, including poor ability for avoiding local optimum and low convergence rate. The definition of the heuristic quasi-oppositional point is given, and its outstanding performance is proven. Then, a heuristic quasi-oppositional searching operator is designed for accelerating convergence rate and enhancing the global exploration ability of the algorithm. Meanwhile, the mutative scale chaos mutation operator based on the piecewise linear chaotic map(PWLCM) mapping function is constructed for boosting the ability of escaping local optimum. Simulation results on benchmark functions show that the proposed algorithm can avoid local optimum effectively, and achieves significant improvements in terms of convergence speed and accuracy.