In order to solve the shortcomings of prediction error method in predi ction accuracy and image imperceptibility,an adaptive selecting block reversible image watermarking algo rithm based on improved pixel-value-ordering (PVO) is proposed.Firstly,the original image is divided into blocks,and the noise leve l of the subblock is calculated using the gradient between the correlation pixel and the mean in the subblock.Then,t he prediction error histogram is obtained by combining the PVO prediction operator.Finally, the embedding position of watermarking is adaptively selected according to the noise level of subblock,an d the watermarking is embedded using the improved PVO algorithm.The algorithm can adaptively select the subblo ck of embedding watermarking according to the characteristics of the original image,improves the pixel predi ction accuracy,and reduces the distortion caused by watermarking embedding.After extracting the watermarking i nformation,the original image can be restored without distortion.The experimental results show that the impro ved strategy of the noise level judgment and PVO algorithm in this paper can effectively improve the performan ce of the reversible image watermarking algorithm.Compared with the related references,the peak signal-t o-noise ratio of the watermarked image is improved by about 1.3% on average in the same embedding capacity,and i t has better imperceptibility.