| 摘要 |
874,986. Electric analogue calculating. AUTOMETRIC CORPORATION. April 15, 1959 [May 15, 1958], No. 12777/59. Class 37. General.-An electric analogue network comprising plural electric junction points interconnectible by conducting branches into a matrix of rows, columns and diagonals includes a resistor between each junction and every junction adjacent thereto along said columns, rows and diagonals; the resistors being of substantially identical values; and adjustable means provided to inject an alternating voltage into selected individual conducting branches with means to measure the voltage between any such junction point and a selected other junction point; and is applicable to error correction computations necessary in mapping terrain by superimposition of successive overlapping photographs of optical or radar images of the surface from an aircraft flying a predetermined course; the relative positions of the individual elements being specified with regard to a known point on one of them in terms of cartesian co-ordinates of corresponding points (e.g. the centres) located in such elements. Since the algebraic sum of the displacements along each separate co-ordinate axis intersecting such points in a closed path is in general not zero due to intrinsic errors of the mapping system, it is necessary to effect for each coordinate an optimum distribution of such closure error among all the displacements of the closed path to ensure optimum distribution of the component photographs and to redistribute the multiple closure errors in plural closed paths selected to cover the area to be mapped, so that each pair of adjacent paths includes two or more over-lapping photographs in common, and corrected co-ordinates for the corresponding points of all photographs of a series may be computed. Further corrections derived from independently known landmarks may be introduced, and the effect of uncertainty in the known positions thereof may be estimated. Fig. 1 shows a series of circular superimposed transparent photographs 1A, 1B, 1C ... 1X; 2A, 2B, 2C ... 2X, &c. of a radar P.P.I. display in an aircraft flying along predetermined parallel courses arranged in rows A, B, C to X and columns 1, 2, 3 to N with a lubber line h indicating the aircraft heading for each exposure; adjacent photographs being matched by coincidence of common portions thereof, and the relative positions of each pair being completely defined in terms of orthogonal linear co-ordinates x, y of corresponding points and angular co-ordinates # of the relative inclination of the lubber lines h. In a pair of overlapping photographs of a series, e.g. 2 and 4 (Fig. 2), with centres 2<1> and 4<1>, check points t1, t2 define a true direction with respect to which the y and x axes are positioned North and East; t1 being of known latitude and longitude. The co-ordinates of 2<1> may be determined by measurement of the differences in x and y between 2<1> and t1 while 41 may be determined from the co-ordinated displacements #x and Ay from 21 and the angle ## between the lubber lines of the pair. Specification 853,794 describes apparatus for deriving the x, y, # values of successive overlapping photographs. Fig. 3 shows a concrete example of the mutual linear spacings of the centres of an array of nX overlapping photographs 1A, 2A, 3A, &c., 1B, 2B, 3B, &c., 1C, 2C, 3C, &c. arranged in n columns 1, 2, 3 ... n and x rows A, B, C ... X as in Fig. 1, wherein due to navigation and timing errors the successive photographs do not lie precisely in a rectangular array; the measured differences #x and Ay between adjacent centres taken both diagonally and along the co-ordinate axes being indicated in magnitude and sense; referred to centre 1A as starting point with the' convention that variations are positive from a point in a row or column of lower order to a point in one of higher order, or vice versa. Due to the errors referred to, algebraic summation of respective co-ordinate variations along any closed path intersecting the points of the array will not result in closure of the path; i.e. the algebraic sum along the whole path will in general not be zero for either co-ordinate. The closure errors are distributed among the measured co-ordinates of the centres of the respective photographs by accumulating the measured values of the spacings of the centres from an origin point along an arbitrary route intersecting all the centres, to assign to each centre and for each co-ordinate an " assumed value " designating such centre. A series of closed paths is notionally traced out among the centres utilizing such assumed values and the closure error for each path is obtained; utilizing an electrical analogue matrix representing the closed paths into each path of which there is introduced a generator whose output is proportional to the closure error in that path for each co-ordinate, whereby the coordinate voltages arising in the analogue circuit corresponding to the points shown in Fig. 3 are adjusted with regard to an origin point (e.g. A1) so as to be corrected for the distributed values of the closure errors for each loop and for each co-ordinate direction. Considering the four photographs 1A, 1B, 2B, 2A, the accumulated closure error for the x co-ordinate around the loop through their centres A1, B1, B2, A2 is +0.990 (Fig. 4), and may be corrected in the corresponding analogue network (Fig. 5) comprising four junction points A1, B1, B2, A2 interconnected by equal resistors R, with an A.C. or D. C. low impedance generator G (A1, A2) in series with the corresponding branch whose voltage is given by where x(A2), x(A1) are the accumulated x co-ordinate values for points A2 and A1 over the path A1, B1, B2, A2, #x(A2A1) is the variation in values of x from point A2 to A1, K is a constant, so that the generator voltages is 0.012 - poled according to the convention that a K, positive voltage requires generator negative connected to A1 and vice versa. In general the value of the closure error generator voltage G(mn+) for two matrix points m, n of assumed x co-ordinates x(m), x(n) by reason of accumulated x co-ordinates from an origin along an arbitrary path through m, n is given by where #x(m,n) is the change in x between m and n measured in direction mn, K is a constant. The voltages of junctions B1, B2, A2 (Fig. 5), are then varied with respect to that of A1 as an algebraic measure of the changes to be made in the x co-ordinate values of points B1, B2' A2 to obtain relaxed values compensating for the x closure error. A similar procedure is utilized for correcting the y co-ordinates. When A.C. generators are utilized, phase sense is used to denote polarity in relation to a reference voltage. Further correction of the assumed co-ordinates may be achieved by developing generator voltages corresponding to closure errors round loops incorporating diagonals B1, A2 and A1, B2 which are inserted into corresponding resistor diagonals of the corresponding matrix (Fig. 6); the generator voltages being given in sense and polarity by equation (2) above; and the corrections being added to the assumed coordinate point in terms of the voltages appearing in Fig. 6 between such points and the origin A1. Structure.-Fig. 8 shows an analogue matrix circuit for obtaining the corrections for application to the x (and similarly to the y) coordinates of the nine points A1, B1, C1, C2, B2, A2, A3, B3, C3 (Fig. 7). A junction point corresponds to each point of Fig. 7 and each such point is connected directly and diagonally to its neighbours by equal resistances R along the rows, columns, and diagonals of the array. For each set of four rectangular junction points there are provided three error voltage generators G disposed respectively in a row (or column) adjacent to a point and in each diagonal adjacent thereto. The error voltage generators are energized with voltages derived according to equation (2) above from comparison of the difference between the assumed values of each pair of adjacent points and the measured separation thereof. Where data is not available for a particular #x (or AY) spacing the corresponding branch is opencircuited. Further corrections may be introduced by reference to check points of known geographical location lying within one or more photographs; e.g. t3 lying within photograph 1A from which the co-ordinates of the centre thereof may be accurately determined, and t'4 t5 lying within photographs 2A, 2B, 3A and 3B, and within 2B, 2C, 3B and 3C respectively. For the x co-ordinate, a generator GC is connected in series with resistor R between the matrix origin A1 and earth to insert a voltage representing the difference between the separation between t3 and the centre Al of photograph 1A and that between the x co-ordinate of t3 and the assumed x co-ordinate of A1, a further generator GC is connected in series with resistor R between matrix point A2 and earth to insert a voltage representing the difference between the separation along the D.C. axis between t4 and the centre A2 of photograph 2A and that between the x co-ordinate of t4 and the assumed x co-ordinate of A2. Similarly check point generators GC and resistors R are connected between a common point representing T5 and junctions A2, B3, C2, C3 representing the centres of four photographs of Fig. 1 on which the check point appears; i.e. the photographs to whose centres the check point is nearest; the inserted voltages being derived for the several photographs in the same manner as for T4. Point T5 of the matrix is earthed through generator GU introducing a voltage + 8x at a frequency differing from that of generators GC to represent uncertainties of the x co-ordinate of the check point T5. Other such generators may be provided operating at different frequencies to represent uncertainties at other check points. The X co-ordinate connections are determined from the matrix |
