In order to solve the speed bottleneck of visual big data processing and computing speed problem, a fuzzy clustering locality preserving projections algorithm is proposed for feature extraction of visual data. Firstly, the image is segmented by some method. Then, the image is characterized by statistical methods to obtain the corresponding visual data. Then, the visual data is extracted by the fuzzy clustering locality preserving projections proposed in this paper. Finally, through AdaBoost, the extracted features are identified and classified. Through a large number of experiments on the internationally published UCAS-AOD and Flower-102 datasets, the experimental results verify the effectiveness of the fuzzy clustering locality preserving projections algorithm in visual data feature extraction.