Coefficient Random Permutation Based Compressed Sensing for Medical Image Compression
Compression of medical data remains challenging because of the loss in clarity of compressed images. In medical field, it is necessary to have high image quality in region of interest. This paper presents a Compressed Sensing (CS) method for the compressi
- PDF / 176,772 Bytes
- 8 Pages / 439.37 x 666.142 pts Page_size
- 59 Downloads / 290 Views
Abstract Compression of medical data remains challenging because of the loss in clarity of compressed images. In medical field, it is necessary to have high image quality in region of interest. This paper presents a Compressed Sensing (CS) method for the compression of medical images. Coefficient random permutation (CRP) based CS is used in this paper for compression of medical images. The different image block has different sparsity. If the nearby pixel values in a block have stronger correlation, then they are strongly sparse, otherwise they are said to be weakly sparse. The main objective of using this method is to provide high quality compressed images thereby maintaining a balanced sparsity of the reconstructed images. As a result performance gain would be high. Experimental results show that CRP based CS helps achieving better PSNR values even with lesser number of measurement values. Keywords CRP matching pursuit
Compressed sensing Gaussian matrix Orthogonal DWT
1 Introduction Medical images are captured and transferred between hospitals for review by physicians. Also these images are required to be stored for the reference of physicians and patients in future. An efficient compression algorithm is required for effective storage of data for future use. Conventional compression (CC) helps in R. Monika (&) S. Dhanalakshmi ECE Department, SRM University, Kattankulathur, Chennai, India e-mail: [email protected] S. Dhanalakshmi e-mail: [email protected] S. Sreejith School of Electrical Engineering, VIT, Vellore, India e-mail: [email protected] © Springer Nature Singapore Pte Ltd. 2018 A. Kalam et al. (eds.), Advances in Electronics, Communication and Computing, Lecture Notes in Electrical Engineering 443, https://doi.org/10.1007/978-981-10-4765-7_56
529
530
R. Monika et al.
reducing the amount of data. However they do not provide significant reduction in data rate. In order to achieve a good compression ratio, compressed sensing techniques can be used. CS achieves good compression. A signal can be acquired at a rate which is much lesser than nyquist sampling rate by using CS. However in Conventional CS Sparsity of all blocks of the image is not even. Hence images are reconstructed with poor quality. Generally smoother region have stronger sparsity while edges have weaker sparsity. Due to imbalance in sparsity few regions (i.e., highly sparse) will be reconstructed well while other regions (i.e., poorly sparse) are poorly reconstructed. This in whole degrades the quality of the reconstructed images. Also conventional CS cannot be applied for medical image compression as it requires good quality reconstruction with balanced sparsity. In order to achieve balanced sparsity in all blocks of the image, CRP is used. Balanced sparsity is achieved by means of random permutation of the coefficients. CRP method in DWT domain helps in improving CS sampling efficiency and performance.
2 Related Works Candes and Wakin [1] explained that certain signals can be recovered successfully from fewer
Data Loading...
