4. Selection of the Remaining Code PassesWe continue to encode the code passes for which the accumulation rate is greater than or equal to the target rate. Using (1), we calculate the R-D slope and the accumulation rate of each code pass. If the R-D slope of this pass is greater than zero and not greater than ��nopt, it should be discarded, and the remaining code passes will be skipped selleck chemicals Afatinib in this code block. Then, the same processing is performed on the next code block.When all of the code blocks have been processed, we continue by performing the PCRD algorithm during Tier-2 encoding. The main principle of the PCRD algorithm is to seek the optimal truncation points of each code block within certain bit rate restrictions to minimize distortion.

Using the above steps, we can truncate a large number of code passes that are skipped during Tier-2 encoding by using the threshold ��nopt. This procedure can greatly reduce the computation and working memory requirements of the image compression. At the same time, the scope of searching for the optimal R-D slope and the optimal truncation point will be narrowed for Tier-2 encoding.3. Simulation Results and DiscussionIn this section, we focus on remote sensing images in the real world and show how our proposed method works.The proposed rate control algorithm is tested using eight test images that are selected randomly from 30 remote sensing images of different sizes. It is implemented on the Jasper software platform [13], which is defined in Part 5 of the JPEG2000 standard.

In each of the images, we use (5, 3) wavelet filters with six-level DWT decomposition with a code block size of 64 �� 64 (the default coding parameters in the Jasper software). The simulation results in terms of PSNR are shown in Table 1. The results obtained from using the standard PCRD method are also shown in Table 1. In Table 1, ��PSNR is defined as��PSNR=PSNRPROPOSED?PSNRPCRD.(3)Table 1PSNR comparison of the proposed algorithm and the PCRD algorithm.Table 1 shows the performance, in terms of PSNR, of the proposed algorithm and the PCRD algorithm. At different bit rates, the PSNR decreases slightly. In most cases, the difference in the PSNR performance is less than 0.1dB, with the largest difference being 0.151dB. The proposed algorithm is slightly better than the PCRD algorithm at some bit rates, with improvement in the range of 0.

002�C0.017dB. Furthermore, we can Dacomitinib see that the PSNR remains unchanged for a bit rate of 1.0bpp.For comparison, Figure 1 shows the average PSNR difference for eight remote sensing images at the different bit rates. From this comparison, we can see that the differences in the average PSNRs are less than 0.04dB. From Table 1 and Figure 1, we can see that there is little loss of image quality. Therefore, the proposed algorithm can achieve good image quality.Figure 1Average PSNR difference between the PCRD and proposed algorithms.