The invention relates to a crowd density analysis method based on unmanned-aerial-vehicle vision and support-vector-machine classification. The method comprises the following steps: firstly, using a camera to photograph a crowd in an area to obtain a color image; and then carrying out a gray-level processing process on the collected crowd image; then carrying out elimination of noises on the image; then using background subtraction operation to obtain a crowd foreground image; and then using a gray-level co-occurrence matrix-based texture analysis method to extract crowd density features, andsending a result thereof to a support-vector-machine classifier, and thus obtaining crowd density. According to the method, effects that the crowd density can be monitored in real time, and an unmanned-aerial-vehicle can be released at a position where monitoring needs to be carried out are realized, and a method of public safety management is further realized.