Local adaptive registration method for construction of pixel level SAR (Synthetic Aperture Radar) image time sequence

The invention relates to a local adaptive registration method for construction of a pixel level SAR (Synthetic Aperture Radar) image time sequence. The method comprises the following steps: extracting homonymous feature point pairs from master and slave images after data preprocessing; calculating a quadratic polynomial parameter by using a least square method and calculating matching overall error; comparing the sizes of the matching overall error and the given threshold value; if the matching overall error is less than or equal to the given threshold value, determining by the quadratic polynomial from the positional relation of the slave image and the master image and finally performing image registration; and on the contrary, if the overall error is greater than the given threshold value, performing error point clustering to acquire a distortion region, taking normal regions of the master and slave images as a pair of new master and slave images and the distortion region as another pair of new master and slave images, repeatedly calculating the quadratic polynomial parameter and the next steps for the two pairs of new master and slave images until the matching overall error of the homonymous feature point pairs of all the new master and slave images is less than the given threshold value, and then performing image registration.