<p>Since many existing steganalytic features are high-dimensional, this increases training and test time of classifiers, even affects the performance of steganographic detection. Aiming at solving the problem, dimensionality reduction of high-dimensional spatial-domain steganalytic features is investigated and a dimensionality reduction algorithm based on PCA and equal interval division is proposed. The proposed algorithm first removes these steganalytic features with all zero values, and adopts PCA to process each submodel respectively and combines the processed features. Then, it employs Pearson correlation coefficient between features and the class label to measure the relevance and removes some features with weak relevance. Following that, it utilizes Pearson correlation coefficient between features and the class label to measure the relevance, and exploits Pearson correlation coefficient between features to measure the redundancy, and uses the method of equal interval division and ranking for feature selection and selects some features ranked in front. For validating the performance, the proposed algorithm is compared with the algorithm based on Fisher score, maxSRM and the features processed by exp-Hellinger. Experimental results show that the proposed algorithm can achieve better dimensionality reduction performance.</p>