A JITTER DETECTION METHOD BASED ON THE INTEGRATION IMAGING MODEL

Abstract. Satellite jitter is a random error source which leads to image degradation. This paper proposes a method to detect the time-variant jitter using multispectral images. In the method, multispectral images are adopted for their large overlap to obtain the parallax map. The imaging process is analyzed in details, and an integration imaging model is constructed, which takes fully into account the time-variant jitter property and builds the relationship between object space with image space. Besides, multispectral images of ZY-3 satellite were used for experiments, and results show that the presented method obtains the jitter curve with the error of amplitude, frequency and phase not more than 0.0591 px, 0.0006 Hz and 0.007 rad, respectively. Results demonstrate the performance of the presented method in jitter detection.