引入拉格朗日算子的纹理深度联合比特分配方法

Joint Bit Allocation Method for Texture Videos and Depth Maps with Lagrange Multiplier

为了提高三维视频合成质量,提出一种新的纹理视频和深度图编码比特分配方法.该方法在比特分配中引入拉格朗日算子参量,从而在限制比特下进一步降低虚拟合成误差.首先依据统计实验并采用最小二乘拟合的方法,对纹理视频和深度图分别建立以拉格朗日算子为参数的编码误差模型和编码比特模型,然后利用编码误差和虚拟合成误差的相关性进一步推导出基于拉格朗日算子的虚拟合成误差的模型,最后结合所推导的模型建立虚拟合成误差最小化目标函数,利用无约束优化方法求解最优的纹理视频和深度图编码拉格朗日算子,并依据此算子进行比特分配.实验结果证明:与固定比特算法和全搜索量化步长算法相比,该方法在不同的编码条件下使合成视频质量平均提高0.25dB和0.68dB,且得到的编码总比特与目标比特最为接近,平均误差仅为0.45%.

A novel joint bit allocation method for the texture videos and depth maps coding is proposed to improve the synthesis quality of 3D video（3DV）.The contribution of the proposed method is that the Lagrange multiplier is introduced into the bit allocation as a parameter to further reduce the virtual synthesis distortion under the constrained bit.Both distortion model and bit model for texture videos and depth maps are constructed through statistical experiments,and the Lagrange multiplier is treated as a parameter in both of the models.Then,a model of warping errors is deduced by a comprehensive analysis on the relationship between the coding distortion and warping error,and an objective function for minimizing the warping error is designed using this model.A constrained optimization method is used to obtain the optimal Lagrange multipliers of the texture video and depth map coding,and then the bit is allocated using the resulting Lagrange multipliers.Experimental results verify that compared with the full search based bit allocation algorithm and fixed bit algorithm,the proposed method achieves an average gain of 0.25 dB and 0.68 dB under different coding conditions,respectively.Meanwhile,the coding bit of the proposed method is closer to the constrained bit,and the average error is only 0.45%.