Due to malfunctioning and mis-calibration of cells in digital x-ray detectors as well as impurities on the scintillator screens, stripe artifacts arise in the sinogram which in turn generate ring artifacts in the reconstructed x-ray computed tomography images. In this paper, a novel technique is proposed for the detection and removal of stripe artifacts in a sinogram with a view to suppress the ring artifacts from the tomographic images. To accurately detect the stripe creating pixels using a derivative-based algorithm, at first the sinogram is windowed to create a sub-sinogram by keeping the pixel of examination at the center position in the sub-sinogram. The other pixels in the sub-sinogram are selected from a polyphase component of the sinogram. A new mathematical index is proposed here to isolate the strong and weak ring-generating stripes from the good ones. For the correction of strong ring artifacts resulting from the defective detector elements and dusty scintillator crystals, 2D variable window moving average and weighted moving average filters are proposed in this work. On the other hand, a conventionally trusted constant bias correction scheme is adopted to correct the responses of the mis-calibrated detector elements. To evaluate and compare the performance of the proposed algorithm, real micro-CT images acquired from two flat panel detectors under different operating conditions are used. Experimental results show that the proposed method can remove ring artifacts more effectively without imparting noticeable distortion in the image as compared to a recently reported technique in the literature.