The main challenges of multi-object tracking are the heavy occlusion between targets and targets, in/out the field of surveillance and the data association between objects and candidate objects, etc. In this paper, we proposed a multi-object tracking algorithm by collaborative multi-feature based on Kalman filter, first, in the detection program, we extract pedestrians in every frame, and we adopt non-maximum suppression to filter the results; then, in the object tracking and data association program, we distribute a Kalman filter for every target, we construct a collaborative model: color feature-based appearance model, texture feature-based appearance model and a spatial distance information model, for data association; finally, we formulate a likelihood function to choose best match for the target and the candidate. The experimental results show that the proposed algorithm performs well while the targets are occluded, in/out the surveillance field and data association.