Computing the Exterior Orientation Elements of Line-Array CCD Images by Improved Two-Step Solution Method

Since the exterior orientation elements of line-array CCD images are highly correlated, normal collinear equations to compute these elements are ill-posed and the error of least square estimation is very large. This paper puts forward an improved two-step solution method to compute the exterior orientation elements based on Tikhonov regularization theory and detection of outliers. The first step of the method is to construct a new regularization matrix based on Tikhonov regularization theory, and is essentially to choose a more appropriate regularization matrix, rather than a traditional unit matrix. While the second step is to use gross error detection to compute the matrix functions of observation value. Experimental results indicated that the method could effectively overcome the correlation problems of exterior orientation elements and achieve high positioning accuracy. For orientation points, the positioning precision is under one pixel; while for checking points, the positioning precision is around one and a half pixels.