An Efficient Hardware Realization of DCT Based Color Image Mosaicing System on FPGA
| Autor: | H. Jayalaxmi, S. Ramachandran |
|---|---|
| Rok vydání: | 2018 |
| Předmět: |
Cross-correlation
Pixel 010505 oceanography Image quality Computer science Color image business.industry ComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION 0211 other engineering and technologies Image registration 02 engineering and technology 01 natural sciences Image (mathematics) Discrete cosine transform Computer vision Artificial intelligence Field-programmable gate array business 021101 geological & geomatics engineering 0105 earth and related environmental sciences |
| Zdroj: | Intelligent Systems in Cybernetics and Automation Control Theory ISBN: 9783030001834 |
| DOI: | 10.1007/978-3-030-00184-1_18 |
| Popis: | Researchers of Surveillance systems in today’s world are exploring various ways to capture continuous image sequences. Capturing these image sequences in a single plane is highly challenging. As a consequence, camera arrays have sprung into action to capture the image sequences in a single plane. This concept of combining the captured images by an array of cameras is referred to as mosaicing. The image mosaicing generally encounters the problem of sacrificing the image quality because of improper mode of image capturing. In order to tackle this problem, Cross correlation (Ccr) method can be implemented. However, it faces quality problems again under brightness variations. This paper introduces a Discrete Cosine Transform (DCT) based image mosaicing method that works effectively against the brightness issues suitable for FPGA implementation. This proposal comprises functional blocks for DCT, inverse DCT, image registration and multipliers and dividers. The image quality assessing parameters like PSNR, MSE, device utilization and timing analysis were considered for performance analysis of the proposed scheme by using Xilinx 14.7 ISE and simulated by using Modelsim6.3 f. The proposed image mosaicing system gives excellent results in terms of image quality (PSNR of over 35 dB) and fast execution time of 23 ms per image of size 1600 × 1200 pixels. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|