Precision Pursuit: A Spectral Decomposition-Driven Adaptive Block Measurement Matrix for Enhanced Compressive Sensing in Imaging

Compressive Sensing (CS) is a relatively new sophisticated technique that finds applications in various fields, and the selection of a measurement matrix influences the effectiveness of CS in image processing by identifying sparse informative pixels for sampling. The randomness in the selection of the measurement matrix results in variations in the assessment values. In this context, the Spectral Decomposition-Driven Adaptive Block Measurement Matrix (SD-DAB) method is proposed to improve the objective evaluation of images. The main aim of the proposed method is to obtain recovered images with the support of speed and quality. To accomplish this, the SD-DAB has been meticulously designed, which employs adaptive block processing, a technique that divides an image into smaller blocks and processes each block individually, adjusting the processing parameters based on the content of each block. This allows for more efficient and targeted analysis or enhancement, as the method adapts to variations in texture, brightness, or other local characteristics of the image. and the processed image matrix is analyzed with spectral decomposition. To evaluate the results, the proposed SD-DAB is contrasted with traditional methods like. The measured matrices were evaluated and compared with image quality and computational time, including the Peak Signal to Noise Ratio (PSNR), Signal to Noise Ratio (SNR), Structural Similarity Index Matrix (SSIM), and Mean Square Error (MSE). The evaluation was accomplished for distinctive sub-rates ranging from 0.1 to 0.9 and with images of varying sizes. The proposed method demonstrates superior performance at sub-rate 0.9 applied to a 512×512 sized woman with a dark hair image. It outperforms other methods with a higher PSNR of 38.49dB, SNR of 32.24dB, SSIM of 0.56, lower MSE of 9.20, and notably faster computational time of 29.61secs