The supervised local structure and div ersity projection (SLSDP) algorithm is easily affected by the parameter settings .In this paper,a parameter-free direct orthogonal discrimi nant local diversity projection analysis algorithm is proposed,which is supervised and parameter-free.The algorithm calcu lates the weights using the cosine distance between samples and dynamically determines neighbors of each data point,so that it doesn′t need to set any parameter.It is a supervised feature extraction algorithm because the class information of sa mples is considered in the calculation of the diversity weights.In order to solve the small sample size problem,a direct metho d is given.The orthogonal projection matrix can be directly extracted from the original high-dimensional space according to the properties of the Laplacian matrix.Experimental results on ORL and AR face databases show that the recognition rates of the proposed method can be increased by 2.3% and 1.31%,respectively,which demonstrates that the proposed method is effective and feasible.