The invention relates to the technical field of the sensor data processing, especially to a distributed coordination algorithm and data fusion mechanism for a sensor network. Sensor-level data fusion and central data fusion are carried out respectively. To be specific, during the sensor-level data fusion process, pretreatment of data collected by a sensor is carried out and different characteristic quantities of a measured system are extracted; and during the central data fusion process, integration of the characteristic quantities, extracted by the sensor-level data fusion process, of the measured system is carried out to obtain accurate state estimation of the measured system. According to the invention, the result of the central data fusion is used for correcting the sensor-level data fusion result, so that problems of low precision, single information, and high data fusion calculation load of the single-sensor system can be solved. The algorithm and the mechanism can be applied to multi-source image integration, a robot and an intelligent instrument system, a battlefield and a pilotless aircraft, an image analysis and understanding, target detection and tracking, and automatic target identification.