The invention provides an image target matching detecting method based on multi-particle swarm optimization. The method comprises the steps of inputting a to-be-detected image and a target image, calculating a common characteristic space between the input to-be-detected image and the target image in a preset manner, and furthermore constructing an energy function which corresponds with the common characteristic space; acquiring each local extreme point on the energy function which corresponds with the common characteristic space based on the multi-particle swarm optimization, wherein the number of the local extreme points are determined by the number of matching detections between the to-be-detected image and the target image; according to each acquired local extreme point, determining an area which corresponds with each local extreme point in the to-be-detected image through the energy function that corresponds with the common characteristic space, and outputting each determined area as a final matching detecting result between the to-be-detected image and the target image. The method has advantages of reducing calculation amount, improving calculation speed and realizing high reliability and high precision in image matching.