FIELD The present invention generally relates to a computer image processing field, particularly to a method for removing streak from detector cell with performance difference. BACKGROUND A computed tomography (CT) is a common endoscopy detection technology in the modern medicine. The CT technology obtains an image of a detected region of a human body by mainly rotating X-ray beams and a detector around the human body, and continuously carrying out section scanning, receiving attenuated X-ray information penetrating the human body by the detector and inputting said information into a computer, reconstructing an image through the computer based on the received attenuated X-ray information during each scanning. The image reconstruction in the CT technology is an important factor that would affect the detection result. How to clearly and accurately reproduce an image of the detected region is a problem that has always been studied by the person skilled in the art. From an imaging principle of a CT system, there are multiple factors that would affect a quality of a CT image.For example, in the existing CT system, due to reasons on the structure and manufacturing process of the detector itself, there would inevitably be some performance difference between detector cells on the detector , i.e., there would be some detector cells with performance difference. Projections sensed by those detector cells with performance difference cannot accurately reflect the image of the detected region, thus cannot be applied to the CT image reconstruction directly. An existing processing method is: in the case that the detector cells with performance difference are determined to be present, usually by an interpolation method, a projection on the detector cell with performance difference is estimated from projections sensed on two detector cells adjacent to the detector cell with performance difference. However, there would be some estimation error between a correct projection of the detector cell with performance difference and the estimated projection. Therefore, during the CT image reconstruction, when an image is reconstructed with projections using a filter backprojection method, the estimation error will be back-projected onto the reconstructed image, thereby forming a streak on the reconstructed image, thus affecting the quality of the CT reconstructed image. Accordingly, it is necessary to provide an improved method for solving at least one of the technical problems as mentioned above. SUMMARY One aspect of the present invention is to provide a method for removing streaks from detector cells with performance difference, the method comprising the steps of: al) estimating a projection on a detector cell with performance difference for each of one or more views; a2) reconstructing one or more initial images with the estimated projection on the detector cell with performance difference for each view; a3) estimating a projection error on the detector cell with performance difference for each view from the reconstructed one or more initial images having streaks; and a4) reconstructing one or more output images with the estimated projection error on the detector cell with performance difference for each view to remove the streaks from the reconstructed one or more output images. In one embodiment of the present invention, in the method as mentioned above, the step al) comprises estimating a projection on a detector cell with performance difference using an interpolation method. In the method as mentioned above, the step a4) comprises: subtracting the estimated projection error on the detector cell with performance difference from the estimated projection on the detector cell with performance difference to obtain a corrected projection on the detector cell with performance difference; and reconstructing one or more output images with the corrected projection on the detector cell with performance difference. In one embodiment of the present invention, in the method as mentioned above, the step a2) comprises reconstructing one or more initial images with the estimated projection on the detector cell with performance difference using a filter backprojection method, and the step a4) comprises reconstructing one or more output images with the estimated projection error on the detector cell with performance difference error using a filter backprojection method. In one embodiment of the present invention, in the method as mentioned above, the step a3) comprises: extracting an image stripe along a projection ray of the detector cell with performance difference from the reconstructed one or more initial images; and estimating the projection error on the detector cell with performance difference using one or more gray values of one or more pixels in the extracted image stripe. In one embodiment of the present invention, in the method as mentioned above, the step a3) further comprises: forming a first straight line between a position of an X-ray source and a position of the detector cell with performance difference; selecting second and third straight lines parallel to the first straight line, wherein a distance from the second straight line to the first straight line is the same as a distance from the third straight line to the first straight line; and estimating the projection error on the detector cell with performance difference using one or more gray values of one or more corresponding pixels of the first, the second and the third straight lines. In one embodiment of the present invention, in the method as mentioned above, the step a3) further comprises: dividing the first straight line into M segments; dividing respectively the second and the third straight lines into corresponding M segments as well; computing the projection error on the detector cell with performance difference using the following equation: wherein Pet11' row,view) represents the projection error on the detector cell with performance difference, IS(K2) represents a gray value of the kttl segment on the first straight line, Is(K.l) represents a gray value of the ktf! segment on the second straight line, IS(K.3) represents a gray value of the kttl segment on the third straight line, M represents the number of the segments divided, ci represents a constant mapping a streak level to the projection error on the detector cell with performance difference, and a represents the number of i row, view) on the detector cell with performance difference, improving quality of CT reconstructed images. Fig. 4 illustrates a flow chart of a method for removing streaks from detector cells with performance difference in accordance with a second embodiment of the present invention. Now referring to Fig. 4, the method for removing streaks from detector cells with performance difference in accordance with a second embodiment of the present invention comprises the following steps: In Step s21, a projection P(n, row, view) 0n a detector cell D(n,row) with performance difference for each of one or more views is estimated, in which 11 is a column number of the detector cell D(n, row) vvith performance difference, row is a row number of the detector cell D(n, row) vvith performance difference, view is a view number of the projection. In the present embodiment, the projection P(n, row, view) on the detector cell D(n,row) wrth performance difference may be estimated using a known interpolation method. The known interpolation method may be any suitable interpolation method. For example, the projection P(n, row, view) on the detector cell with performance difference is estimated with the projections sensed on two detector cells adjacent to the detector cell D(n, row) wim performance difference by a known interpolation method. In Step s22, one or more initial images I are reconstructed with the projection P(n, row, view) on the detector cell with performance difference for each view estimated through Step si 1, each of the reconstructed one or more initial images I being an N X N image. In the present embodiment, one or more initial images I may be reconstructed with the estimated projection P(n, row, view) on the detector cell with performance difference using a filter backprojection method. Since there is an estimation error ^0- row, view) between the estimated projection P(n, row, view) 0f the detector cell D(ii,row) with performance difference in Step s21 and a correct projection of the detector cell D(n, row) vvith performance difference, there exist streaks in the reconstructed one or more initial images I in Step s22. The streaks present on the reconstructed one or more initial images I are generally distributed along a straight line, while there is very little straight line structure in images of a human tissue. Therefore, if a straight line structure can be identified from the reconstructed one or more initial images I and pixels on said straight line can be smoothed, then streaks can be removed from the reconstructed one or more initial images I. Accordingly, the following steps are to remove the streaks on the reconstructed one or more initial images I in light of such ideas of invention. In Step s23, a possible streak direction for each of one or more pixels in the reconstructed one or more initial images I is calculated. Referring to Fig. 5, the position of the X-ray source S and each pixel P(i, j) of one or more pixels in the reconstructed one or more initial images I form a first straight line SP, wherein i=[0, N], j=[0, N]. The first straight line SP defines the possible streak direction for each pixel P(i, j) of one or more pixels in the reconstructed one or more initial images I, i.e., the possible streak direction is a direction along the first straight line SP. In Step s24, a streak feature for each pixel P(i, j) along the possible streak direction SP is identified. Continuing to refer to Fig. 5, first, several pixel points PQi, , PQR adjacent to the pixel P(i, j) are selected along the possible streak direction SP in the reconstructed one or more initial images I. A straight line is selected to pass each pixel point PQk (k=[l, R]) of the several pixel points and perpendicular to the possible streak direction SP, which is referred to as a second straight line LI. For each pixel point PQk of the several pixel points, find two pixel points PQki, PQk2 adjacent to the pixel point PQk in the second straight line LI, compare a gray value of the pixel point PQk to gray values of its adjacent two pixel points PQki, PQk2, and identify a streak feature for the pixel P(i, j) using the compared results. In Step s25, identify whether a pixel P(i, j) is a streak pixel? In the case of identifying that the pixel P(i, j) is a streak pixel, the process proceeds to Step s25; in the case of identifying that the pixel P(i, j) is not a streak pixel, the process returns back to Step s24. In one embodiment of the present invention, if there is always the gray value of the pixel point PQk that is smaller than the gray values of its adjacent two pixel points PQki> PQk2for each pixel point PQk, k=[l, R] of the several pixel points, the pixel P(i, j) is regarded as a streak pixel. Or, alternatively, in another embodiment of the present invention, if there is always the gray value of the pixel point PQk that is bigger than the gray values of its adjacent two pixel points PQki, PQk2for each pixel point PQk,k=[l, R] of the several pixel points, the pixel P(i, j) is regarded as a streak pixel. In Step s26, when a pixel P(i, j) is identified as a streak pixel, the streak pixel P(i, j) in the streak direction is smoothed so as to smooth a difference image. Referring to Fig. 5, similarly, a second straight line LI is selected in a direction passing the streak pixel P(i, j) and perpendicular to the streak direction SP. Then, two pixels PI, P2 adjacent to the streak pixel P(i, j) are found in the second straight line LI. In one embodiment of the present invention, in Step s26, the steak pixel P(i, j) in the streak direction is smoothed using the following equation: I(i,j) = wlXIl + w2 Xl(i,j) + w3 XI2 (4) wherein wl , w2 and w3 represent weighting factors, iG.j) represents a gray value of the streak pixel pGj), H and I2 respectively represent gray values of the two pixels PI, P2 adjacent to the streak pixel pGj), and KiJ) represents a correct gray value of the streak pixel pG, J). The positions of each pixel point PQk and its adjacent two pixel points PQki, PQk2 and two pixels PI, P2 adjacent to the streak pixel pGj) may be determined based on a distance from the X-ray source S to the streak pixel p0- j), a fan angle 1 (an angle between the connecting line SP from the X-ray source S to the streak pixel pG> j) and the connecting line SO from the X-ray source S to its rotating center O) of the X-ray source S and a projection angle $ (an angle between the connecting line SP from the X-ray source S to the streak pixel pG, j) and the positive direction of the horizontal x-axis) of the X-ray source S and the like. The individual gray values may be determined based on the individual positions in the reconstructed one or more initial images I. A difference image ID is obtained by the following equation: ID = 11 (5) wherein I represents a reconstructed initial image, I represents a corrected image obtained from IG, j). After obtaining the difference image ID , the difference image ID is smoothed to remove a high-frequency component from the difference image ID . In Step s27, one or more output images h are updated with the smoothed difference image, such that the output images I0 will become more natural, without flicker. In one embodiment of the present invention, the output image I0 is obtained using the following equation: I„ = I -I- C2 x smooth(T — i) (g) wherein smooth represents a smooth operator and c2 represents a parameter for mapping a difference level to the difference image (I-0 after the smooth operator. The method for removing streaks from detector cells with performance difference of the second embodiment of the present invention can remove the streaks from the reconstructed one or more initial images I directly by identifying pixels in the streak direction from the reconstructed one or more initial images I having streaks and smoothing the streak pixels, thus improving quality of CT reconstructed images. The projections on the detector cells D(n, row) vvith performance difference may be estimated using the projections on the adjacent detector cells by a known interpolation method. However, tiny structures in the human body may cause a larger estimation error, which cannot be reduced by an advanced interpolation method. Therefore, the present invention provides a method for removing streaks from detector cells with performance difference of the third embodiment. The reason for causing said estimation error is a low sampling rate due to the detector cell D(n, row) vvith performance difference. In the present method, the estimation error may be reduced using redundant information from a sine diagram. Since the conjugate projection ray of the X-ray source S is most likely to pass similar tiny structures, and since a taper angle of the beam 162 of X-rays cannot use a conjugate projection of the conjugate projection ray directly, the present method utilizes a high sampling rate of a conjugate detector cell at a conjugate position to generate correct date, thereby obtaining the estimation error caused by the low sampling rate of the detector cell D(n' row) vvith performance difference by an interpolation method. the method for removing streaks from detector cells with performance difference of the third embodiment of the present invention is described in detail in the following. Fig. 6 illustrates a flow chart of the method for removing streaks from detector cells with performance difference in accordance with the third embodiment of the present invention. Now referring to Fig. 6, the method for removing streaks from detector cells with performance difference in accordance with the third embodiment of the present invention comprises the following steps: In Step s31, a projection P(a row,view) 0n a detector cell Dfarow) with performance difference for each of one or more views when the X-ray source S is located in each of one or more positions is estimated. When the rotatable gantry 12 is located in a certain scanning position, the X-ray source S is located in a position as shown in Fig. 7a, in which projections on detector cells adjacent to the detector cell D fa row) vvith performance difference may be calculated from the sensed information on the detector cells adjacent to the detector cell D(n, row) vvith performance difference. In one embodiment of the present invention, by the sensed information on four detector cells adjacent to the detector cell D(n, row) vvith performance difference, projections P(n ~ 2,row_2jview) P(n - 1, row^, view) P(n + i,row+1,view) and +2' on the four detector cells adjacent to the detector cell D(n, row) vvith performance difference in the position may be calculated respectively. According to different scanning methods, the row numbers 10W-2^ _1, and +z of the adjacent four detector cells may be the same as or different from the row number row of the detector cell D(n, row) with performance difference. Then, using any suitable known interpolation method, the projection P(n, row, view) on the detector cell D(n, row) vvith performance difference may be estimated with the calculated projections P(n-2,raw_2,view) ^ P(n-l,row_-,,view) P(n + 1, row+1, view) and P(n + 2, row+2, view) detector cells adjacent to the detector cell D(n' row) vvith performance difference, i.e., as shown in the following equation: However, the present invention is not limited thereto. In another embodiment of the present invention, the projection P(n, row, view) on the detector cell D fa row) vvith performance difference may be estimated with the projections on two or more detector cells adjacent to the detector cell D(n, row) with performance difference. In Step s32, a projection error on a conjugate detector cell for a conjugate view when the X-ray source S is located in a conjugate position relative to each position is estimated, wherein each position and its conjugate position are located in a connecting line between the X-ray source S and the detector cell Dfa row) with performance difference when the X-ray source S is located in each position. As the rotatable gantry 12 rotates, the X-ray source S and the detector 18 rotate around the rotating center O along with it. When the rotatable gantry 12 rotates to a conjugate position of the position of the X-ray source S as shown in Fig. 7a, i.e., the X-ray source S rotates to be located in the conjugate position as shown in Fig. 7b along with it, i.e., the conjugate position of the X-ray source S as shown in Fig. 7b corresponds to the position of the detector cell Dfa, row) with performance difference when in the position of the X-ray source S as shown in Fig. 7a. When in the conjugate position of the X-ray source S as shown in Fig. 7b, the detector cell D(n, row) vvith performance difference rotates from the position as shown in Fig. 7a to the position as shown in Fig. 7b, and at this moment, the conjugate detector cell of the detector cell Dfa, row) vvith performance difference just corresponds to the position of the X-ray source S as shown in Fig. 7a. In one embodiment of the present invention, Step s32 comprises: „_n . . . P-nfn-.row-, view,.) , In Step s321, a conjugate projection cu ■ c c c on the conjugate detector cell uc(,nc- *owc)for a conjugate view when the X-ray source S is located in a conjugate position relative to each position is calculated. In the conjugate position as shown in Fig. 7b, the conjugate projection rays emitted from the X-ray source S will be sensed by the conjugate detector cell Dc(jic,rowcJ 0f the detector cell D(n, row) with performance difference, and , • Prn(rir>row.,,viewf) ... .. the conjugate projection LU c c on the conjugate detector cell Dclnc'rowcJ) may ^e calculated directly from the sensed information of the conjugate detector cell Dc^llc' rowc). In Step s322, a conjugate projection Pc(nc< rowc*viewj Qn the conjugate detector cell uc(.nc-10Wc) for a conjugate view when the X-ray source S is located in a conjugate position relative to each position is estimated. When the X-ray source S is in the conjugate position as shown in Fig. 7b, the projections on the detector cells adjacent to the conjugate detector cell D(n, row) may be calculated by the sensed information of the detector cell in the conjugate position corresponding to the detector cells adjacent to the detector cell D(n, row) vvith performance difference. In one embodiment of the present invention, by the four detector cells adjacent to the conjugate detector cell D(n,row) , projections P«*(nc - 2,rowc_2,viewc) Pc0(nc - 1, rowc-i, viewc) Pc0 (nc + 1, rowc+1. viewj ^ Pc0(nc + 2,rowc+2,viewc) Qn the four detector cells adjacent to the conjugate detector cell D(n' row) in the position may be computed respectively. According row ^^^^c—l TOW to different scanning methods, the row numbers c~2, , c+1 and c+z of the adjacent four detector cells may be the same as or different from the row number rowc of the conjugate detector cell uc^nc-rowc.J Then, using any suitable known interpolation method, the conjugate projection PcCncrowc»viewc) 0n the conjugate detector cell Dc(iic,rowc) may be estimated with the computed projections c0^ c ' c~2' cPc0 (nc - 1, rowc_lP viewc) PcD (nc + 1, rowc+lP viewc) and Pc0(nc + 2, rowc+2,viewc) Qn the detector cells adjacent to the conjugate detector cell uc(.nc- rowc) j e ^ as ghown m fae following equation: Pc(nc, rowc, viewc) = L(Pc0(nc — 2Jrowc_2,viewc),P(.(){nf — l,rowc_1,viewc),Pc ;ij(nc + 1, rowc+1, view,,), Pc0(nr + 2,row(.+2,vie^vc)) (8) However, the present invention is not limited thereto. In another embodiment of the present invention, the conjugate projection Pc(nc,rowc,viewc) Qn the conjugate detector cell Dc(nc,rowc) may be estimated with the projections on two or more detector cells adjacent to the conjugate detector cell uc(.no 10Wc) by a known interpolation method. T o ^^ , , • Prn(iV.rowr, view_) In Step s323, the computed conjugate projection cu c c c on the conjugate detector cell uc(.nc-rowcJ [s subtracted from the estimated P (n row view 1 1 • 1 n conjugate projection c< c> c> cv on the conjugate detector cell Dc(nc,rowc) t0 estjmate the projection error Pee Ovow, view) on the conjugate detector cell LJc(.11cirow'c)) i e ^ as shown in the following equation: pce(n, row, view) = Pc(nc,rowc,viewc) - Pc0(nc,rowc, viewc) (9) In Step s33, one or more output images are reconstructed using the estimated projection P(a row,view) 0n the detector cell D(n,row) with performance difference and the estimated projection error Pcelnc! owc, viewcj on the conjugate detector cell Dc(nci rowc) Furthermore, Step s33 comprises: Step s331, subtracting the estimated projection error pce(nc,rowc,viewc) on the conjugate detector cell Dc(nClrowc) from the estimated projection P(n, row,view) on the detector cell D(n,row) with performance difference to obtain an improved projection P(n, row, view) \ e^ as shown in the following equation: Step s332, reconstructing one or more output images with the improved projection. The method for removing streaks from detector cells with performance difference of the third embodiment of the present invention can remove the streaks from the detector cells D(n, row) vvith performance difference from the reconstructed one or more output images I0by utilizing a conjugate projection ray of the X-ray source S and utilizing a conjugate projection ray sensed by a conjugate detector cell Dc(nc,rowc) tQ the detector cell D(n,row) with performance difference, estimating an estimation error that is possible to occur on the detector cell D(n, row) vvith performance difference with a projection error Pcet,ncrowc-viewcJ that is possible to occur on the conjugate detector cell ^cl^ci rowc.) reducing the estimation error caused by low sampling of the detector cell Dfa row) vvith performance difference, and reconstructing one or more output images with the improved projection P(n, row, view) ^ thus improving quality of CT reconstructed images. Although the present invention has been set forth in details in combination with specific embodiments, the person skilled in the art shall be understood that many modifications and variations may be made to the present invention. Therefore, it should be recognized that the intention of the claims is to cover all these modifications and variations within the real concept and range of the present invention. WE CLAIM: 1. A method for removing streaks from detector cells with performance difference, comprising the steps of: al) estimating a projection on a detector cell with performance difference for each of one or more views; a2) reconstructing one or more initial images with the estimated projection on the detector cell with performance difference for each view; a3) estimating a projection error on the detector cell with performance difference for each view from the reconstructed one or more initial images having streaks; and a4) reconstructing one or more output images with the estimated projection error on the detector cell with performance difference for each view to remove the streaks from the reconstructed one or more output images. 2. The method of Claim 1, wherein the step al) comprises estimating a projection on a detector cell with performance difference using an interpolation method. 3. The method of Claim 1, wherein the step a4) comprises: subtracting the estimated projection error on the detector cell with performance difference from the estimated projection on the detector cell with performance difference to obtain a corrected projection on the detector cell with performance difference; and reconstructing one or more output images with the corrected projection on the detector cell with performance difference. 4. The method of Claim 3, wherein the step a2) comprises reconstructing one or more initial images with the estimated projection on the detector cell with performance difference using a filter backprojection method, and the step a4) comprises reconstructing one or more output images with the estimated projection error on the detector cell with performance difference error using a filter backprojection method. 5. The method of Claim 4, wherein the step a3) comprises: extracting an image stripe along a projection ray of the detector cell with performance difference from the reconstructed one or more initial images; and estimating the projection error on the detector cell with performance difference using one or more gray values of one or more pixels in the extracted image stripe. 6. The method of Claim 5, wherein the step a3) further comprises: forming a first straight line between a position of an X-ray source and a position of the detector cell with performance difference; selecting second and third straight lines parallel to the first straight line, wherein a distance from the second straight line to the first straight line is the same as a distance from the third straight line to the first straight line; and estimating the projection error on the detector cell with performance difference using one or more gray values of one or more corresponding pixels of the first, the second and the third straight lines. 7. The method of Claim 6, wherein the step a3) further comprises: dividing the first straight line into M segments; dividing respectively the second and the third straight lines into corresponding M segments as well; computing the projection error on the detector cell with performance difference using the following equation: wherein Pefa row,view) represents the projection error on the detector cell with performance difference, Isfc'2) represents a gray value of the ^ segment on the first straight line, Is(K.l) represents a gray value of the ^ segment on the second straight line, IS(K 3) represents a gray value of the ^ segment on the third straight line, M represents the number of the segments divided, ci represents a constant mapping a streak level to the projection error on the detector cell with performance difference, and a represents the number of non-zero elements in whereii 8. The method of Claim 7, wherein the step a3) further comprises: determining whether the ^ segment on individual of the first, the second and the third straight lines is in a pixel range of the reconstructed one or more initial images; when it is determined that the ^ segment on individual of the first, the second and the third straight lines is not in the pixel range of the reconstructed one or more initial images, the gray value of the ^ segment on individual of the first, the second and the third straight lines is zero; and when it is determined that the ^ segment on individual of the first, the second and the third straight lines is in the pixel range of the reconstructed one or more initial images, the gray value of the ^ segment on individual of the first, the second and the third straight lines is interpolated from the reconstructed one or more initial images. 9. The method of Claim 6, wherein the distance from the second straight line to the first straight line and the distance from the third straight line to the first straight line are determined by a size of a detector cell of a detector, a distance from the X-ray source to the detector, a distance from the X-ray source to a rotating center of the X-ray source and a reconstruction convolution kernel. 10. A method for removing streaks from detector cells with performance difference, comprising the steps of: bl) estimating a projection on a detector cell with performance difference for each of one or more views; b2) reconstructing one or more initial images with the estimated projection on the detector cell with performance difference for each view; b3) computing a possible streak direction for each of one or more pixels in the reconstructed one or more initial images; b4) identifying a streak feature for each pixel along the possible streak direction; b5) when identifying a pixel is a streak pixel, smoothing the streak pixel in the streak direction so as to smooth a difference image; and b6) updating one or more output images with the smoothed difference image. 11. The method of Claim 10, wherein the step b3) comprises: forming a first straight line by an X-ray source and each of one or more pixels in the reconstructed one or more initial images; and defining the possible streak direction for each of one or more pixels in the reconstructed one or more initial images by the first straight line. 12. The method of Claim 10, wherein the step b4) comprises: selecting several pixel points near the pixel along the possible streak direction in the reconstructed one or more initial images; selecting a second straight line passing each of the several pixel points and perpendicular to the possible streak direction; for each of the several pixel points, finding two pixel points adjacent to the pixel point in the second straight line; comparing a gray value of the pixel point to gray values of its adjacent two pixel points; and identifying a streak feature for the pixel using the compared result. 13. The method of Claim 12, wherein the step b4) further comprises: if the gray value of the pixel point is always smaller than the gray values of its adjacent two pixel points for each of the several pixel points, regarding the pixel as a streak pixel; or if the gray value of the pixel point is always bigger than the gray values of its adjacent two pixel points for each of the several pixel points, regarding the pixel as a streak pixel. 14. The method of Claim 10, wherein the step b5) comprises: smoothing the streak pixel in the streak direction using the following equation: I(i,j) = wlXU + w2 Xl(i,j) + W3 XI2 wherein wl , w2 and w3 represent weighting factors, each of the reconstructed one or more initial images is an N X N image, KM) represents a gray value of the streak pixel p0,j), :J — Lu- ^J II and I2 respectively represent gray values of the two pixels adjacent to the streak pixel p0j) in a second straight line passing the streak pixel p(i' J) and perpendicular to the streak direction, and KM) represents a correct gray value of the streak pixel p0'j). 15. The method of Claim 14, wherein the step b6) comprises: obtaining an output image using the following equation: I0 = I + C2 x smooth(l - i) wherein I0 represents an output image, I represents a reconstructed initial image, I represents a corrected image obtained from KM), smooth represents a smooth operator and c2 represents a parameter for mapping a difference level to ff ~~ 0 after the smooth operator. 16. A method for removing streaks from detector cells with performance difference, comprising the steps of: cl) estimating a projection on a detector cell with performance difference for each of one or more views when an X-ray source is located in each of one or more positions; c2) estimating a projection error on a conjugate detector cell for a conjugate view when the X-ray source is located in a conjugate position relative to the each position, wherein the each position and its conjugate position are located in a connecting line between the X-ray source and the detector cell with performance difference when the X-ray source is located in the each position; and c3) reconstructing one or more output images using the estimated projection on the detector cell with performance difference and the estimated projection error on the conjugate detector cell. 17. The method of Claim 16, wherein the step cl) comprises: computing projections on detector cells adjacent to the detector cell with performance difference; and estimating a projection on the detector cell with performance difference with the computed projections on the detector cells adjacent to the detector cell with performance difference using an interpolation method. 18. The method of Claim 16, wherein the step c2) comprises: c21) computing a conjugate projection pco on the conjugate detector cell for a conjugate view when the X-ray source is located in a conjugate position relative to the each position; c22) estimating a conjugate projection Pc on the conjugate detector cell for the conjugate view when the X-ray source is located in the conjugate position relative to the each position; and c23) subtracting the computed conjugate projection on the conjugate detector cell from the estimated conjugate projection on the conjugate detector cell to estimate the projection error, pc ~~ pco 0n the conjugate detector cell. 19. The method of Claim 18, wherein the step c22) further comprises: computing projections on corresponding conjugate positions of detector cells adjacent to the detector cell with performance difference; and estimating the conjugate projection on the conjugate detector cell with the computed projections on the corresponding conjugate positions of detector cells adjacent to the detector cell with performance difference using an interpolation method. 20. The method of Claim 16, wherein the step c3) comprises: subtracting the estimated projection error on the conjugate detector cell from the estimated projection on the detector cell with performance difference to obtain an improved projection on the detector cell with performance difference; and reconstructing one or more output images with the improved projection on the detector cell with performance difference.