Improved pixel value ordering utilizes two embedding procedures. These are the maximum and minimum embedding procedures. The maximum embedding procedure modifies a pixel by either 0 or 1, while the minimum embedding procedure modifies a pixel by either 0 or − 1. Our counter-embedding strategy modifies a corner pixel using both embedding procedures. This strategy exploits the opposing distortions of each embedding procedure and can embed up to 2 secret bits into a single pixel. The proposed scheme consists of two stages. The first stage applies the counter-embedding strategy to two corner pixels of an input block. Thus, it can conceal up to four bits per block. However, this stage generates some auxiliary information which is embedded in the second stage using the traditional IPVO. The second stage minimizes distortions by only modifying pixels that the first stage ignored. Our experimental results confirm that the proposed scheme improves the visual quality of the recent related works. On average, for embedding capacities of 10,000, 20,000, 30,000, and 40,000 bits, it increases the visual quality by at least 1.2 dB.

改进的像素值排序利用了两个嵌入过程。这些是最大和最小嵌入过程。最大嵌入过程将像素修改为0或1，而最小嵌入过程将像素修改为0或−1。我们的反嵌入策略使用这两个嵌入过程来修改拐角像素。这种策略利用了每个嵌入过程的相反失真，可以将最多2个秘密位嵌入单个像素中。拟议的方案包括两个阶段。第一阶段将反嵌入策略应用于输入块的两个角像素。因此，每个块最多可以隐藏四个位。但是，此阶段会生成一些辅助信息，这些信息将使用传统IPVO嵌入到第二阶段中。第二阶段通过仅修改第一阶段忽略的像素来使失真最小化。我们的实验结果证实，所提出的方案提高了近期相关作品的视觉质量。平均而言，对于10,000、20,000、30,000和40,000位的嵌入容量，它可以将视觉质量至少提高1.2 dB。