Search

A quadrilateral filtering algorithm is proposed in this letter to reduce noise in video synthetic aperture radar (video SAR). The novel filtering approach draws inspiration from the traditional bilateral filter and effectively exploits the similarities in both greyscale and geometric levels. Moreover, the proposed quadrilateral filter incorporates the similarity of time domain and rank-ordered absolute difference (ROAD) statistic for detecting the abundant speckle and impulse noise in video SAR frames. An impulse detector, which is related to the distribution of ROAD, is added in the filtering process to remove the outliers in the intensity domain after time redundancy processing with the target frame image. The proposed quadrilateral filter can obviously smooth various sources of noise, including strong speckle noise and impulse noise, while considering the details of each image frame of the video SAR. The new algorithm has been verified to achieve superior filtering performance with equivalent number of looks (ENL) values increased by 25% on average comparing with other widely used algorithms. [ABSTRACT FROM AUTHOR]