Scholarlink

<p>This work presents an image array sensor that uses compressive sensing (CS) technique to compress image information. First, the impact of five design parameters -the measurement matrix (random or deterministic), image partitioning approach, size of the partitions, compression rate, and the level of sparsity in the image data- are examined. Then, a novel image sensor circuit featuring two operational modes: normal mode and compressive sensing mode is presented. In the compressive mode, the image information is compressed using a deterministic measurement matrix. The design takes advantage of the high correlation of image data in the consecutive frames of the moving objects can effectively compress the difference image information. Consequently, the imager can operate in four different conditions: normal mode with and without compression, and difference mode with and without compression. The results of the compressed normal and difference images show peak signal-to-noise ratio (PSNR) values of 26.11dB and 34.45dB, respectively. The uncompressed normal and difference images yield PSNR values of 29.62dB and 36.89dB, respectively.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

A Multi-Mode Energy-Efficient CMOS Image Sensor Based on Compressive Sensing Technique

  • 期刊论文

  • 发表日期：

    2025年9月29日

查看全文
  • Zahra Sepehri,
  • Sayed Masoud Sayedi,
  • Ehsan Yazdian

摘要

This work presents an image array sensor that uses compressive sensing (CS) technique to compress image information. First, the impact of five design parameters -the measurement matrix (random or deterministic), image partitioning approach, size of the partitions, compression rate, and the level of sparsity in the image data- are examined. Then, a novel image sensor circuit featuring two operational modes: normal mode and compressive sensing mode is presented. In the compressive mode, the image information is compressed using a deterministic measurement matrix. The design takes advantage of the high correlation of image data in the consecutive frames of the moving objects can effectively compress the difference image information. Consequently, the imager can operate in four different conditions: normal mode with and without compression, and difference mode with and without compression. The results of the compressed normal and difference images show peak signal-to-noise ratio (PSNR) values of 26.11dB and 34.45dB, respectively. The uncompressed normal and difference images yield PSNR values of 29.62dB and 36.89dB, respectively.