Research on Digital TDI Technology for Optimizing Sequence Remote Sensing Images Applied in an Imager with Area Array CMOS Sensor

In recent years, the industry of small remote sensing satellites with low-cost area array CMOS imaging sensors prosper, and the trend present rising. Push-frame is the suitable working mode of the satellite with an area array imaging sensor. In order to avoid motion-blur, the situation that the integration time is insufficient of this working mode become inevitable. A digital TDI technology based on the post processing of the ground station is proposed in this paper to solve the problem above. At first, histogram equalization algorithm is conducted to the single frame image sequence with low SNR (Signal to Noise Ratio) and high overlap rate in objects on ground in order to amplify the absolute value of gray scale gradient of each point in the image along each direction; and then, SIFT (Scale Invariant Feature Transformation) algorithm is adopted to extract the feature points of each single frame image in succession; feature points matching between each single frame and forecasted standard image shall also be conducted and the error matching shall be eliminated with the RANSAC (Random Sample Consensus) algorithm to improve the registration accuracy. At last, a transformational matrix shall be calculated with the sequence image according to the coordinate relations obtained after the matching; moreover, multinomial matching registration for image fusion in order to obtain a fused image whose SNR is greatly improved. Experiments show that the matching method is accurate, the brightness and SNR of fused images were significantly improved.