We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
CFA image compression using an efficient cascaded overlapping color transformation.
- Authors
Mohebbian, Mohammad Reza; Chafjiri, Fatemeh Sedighipour; Vedaei, Seyed Shahim; Wahid, Khan A.
- Abstract
Bayer Color-Filter-Array (CFA) is commonly used in single-chip digital image sensors as it efficiently approximates the physiology of the human eye. These sensors capture only one color component per pixel, not an entire RGB image. Conventionally, a Bayer CFA image is converted to an RGB image, and then compressed for processing and storage. However, storing a Bayer CFA image before converting to RGB can preserve more details, which is preferable for post-processing tasks. Due to the arrangement of color channels in a CFA image, compression algorithms cannot efficiently reduce the size of the raw CFA image. In this paper, a Cascade Overlapping Color Transformation (COCT) method is proposed and implemented. In the first step, the Haar Wavelet Transform (HWT) is performed on the image to reduce the spectral redundancy. The second step of the color transformation reduces the redundancy of color components even higher, resulting in a color map that can be compressed better using an image coding system. Experiments performed on several datasets showed that the COCT improves the compression ratio compared to the state-of-the-art techniques. The proposed method improved the compression ratio by 17.6% and 15.6% for 16 bpp and 8 bpp CFA images, respectively. Computational load has also been improved by the proposed method. In addition, the hardware implementation is discussed using a CMOS 65 nm technology that costs 3.18 K gates and 2.05 mW of power.
- Subjects
IMAGE compression; IMAGE sensors; IMAGING systems; WAVELET transforms; EYE physiology; HUMAN physiology
- Publication
Multimedia Tools & Applications, 2023, Vol 82, Issue 28, p43233
- ISSN
1380-7501
- Publication type
Article
- DOI
10.1007/s11042-023-15352-7