In Mars landing mission, sequence images of the parachute opening were obtained during the entry, descent and landing (EDL) process. In the process of parachute tracking, due to multi-scale and multi-view imaging, the moving target in the scene not only has movement changes in the horizontal and vertical directions, but also has spatial rotation and non-coplanar movement, resulted in the rotation, scale and affine transformations of the target in the sequence images. To perform the parachute tracking in the image sequences, an enhanced Camshift algorithm was proposed. This algorithm firstly extracts the color feature as the initial value, and uses Camshift algorithm iteration to automatically adjust the target tracking search for initial target position. Then a BRISKF-Harris feature detector was introduced to relocate tracking area for the sake of target occlusion and nonlinear changes caused by rotation, scale and illumination variation in the tracking process. Use Hamming distance for feature matching, and RANSAC algorithm to eliminate mismatches. The algorithm is verified by the airdrop test images of the Mars probe, and the simulation results show that the algorithm has strong robustness, which can automatically adjust the search window scale factor according to the feature points, and the tracking speed meets the requirements of parachute image frame rate.