基于小波域压缩的图像在传输发生错误时，会引起部分空间区域图像质量的下降。解决这一问题的传统方法是空间域插值或小波域估计。由于传统方法重构质量仍有待提高，为了获得质量更好的重构图像，提出了一种小波一空间域联合估计的差错隐藏算法。该算法首先对小波域低频系数进行简单的八邻域平均来得到初始重构图像；接着在空间域对核心受损区域进行基于块的匹配搜索，通过选择对已知正确小波系数影响最小的块作为匹配块来得到匹配图像；最后将经过滤波处理后的匹配图像变换到小波域，通过与小波域已知正确信息相融合来得到最终重构图像。仿真试验表明，在多数应用场合，该方法与简单的小波八邻域平均低频重构图像相比，客观质量峰值信噪比提高0．5—1dB左右，主观质量也有明显提高。

If error occurs when wavelet-coded images are transmitted over lossy packet networks, some part of the images will suffer from quality degradation. Traditional image reconstruction algorithms can be cataloged into space domain interpolation and wavelet domain estimation. In this paper, hybrid error concealment on both domains is presented, First, a simple reconstruction is carried out on wavelet domain by estimating the lost coefficients by its mean value of the surroundings in the low frequency subband. Second, we search for the match block of the kernel destroyed area according to the known correct wavelet coefficients. Finally, the image is re-transformed to wavelet domain after filtering, and the final reconstructed image is generated by combining both re-transformed coefficients and the original received coefficients. Experiments show that in most cases this method is better than the simple estimation by the mean value of low subband. PSNR is optimized by about 0.5 to 1 dB and the subjective image quality is also improved.