Image reconstruction method and system of compressed sensing lens based on coding aperture, and medium

By setting a coded aperture mask with random patterns in the optical path and combining it with a deep learning reconstruction network, the problem of image quality degradation in low-light and high-frame-rate scenes of traditional imaging systems is solved, and efficient image reconstruction and perception quality improvement are achieved.

CN122265066APending Publication Date: 2026-06-23HANGZHOU HUICUI INTELLIGENT TECH CO LTD
0 Cites 0 Cited by

Patent Information

Application Number
CN202610363470.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-03-24
Publication Date
2026-06-23

AI Technical Summary

Technical Problem

Traditional imaging systems are limited by hardware noise, sampling delay and storage bandwidth in low-light environments, weak signal observation or high frame rate video acquisition scenarios, resulting in degraded image quality. Furthermore, existing compressed sensing methods have high computational complexity and poor real-time performance.

Method used

By employing a compressed sensing lens based on coded aperture, projection modulation is achieved by setting a coded aperture mask with random patterns in the optical path. Combined with a deep learning reconstruction network and a multi-head attention mechanism, physical compression and information recovery are realized.

Benefits of technology

It significantly reduces the amount of data collected, improves image reconstruction efficiency and perception quality, adapts to multiple scenarios, and is suitable for video surveillance and high-resolution medical microscopic image reconstruction under low light conditions.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention provides an image reconstruction method and system of a compressed sensing lens based on a coding aperture, and a medium, and the method comprises the steps: obtaining an original scene image, and carrying out the projection modulation, and obtaining an image received by a sensor; constructing a sensing matrix and a projection matrix, mapping a transfer function of the mask based on the sensing matrix, and performing spatial coding sampling on an image received by the sensor through the projection matrix to obtain a compression measurement result; constructing a deep learning reconstruction network, and processing the compression measurement result based on the deep learning reconstruction network to obtain a two-dimensional feature map; capturing different subspace information for the two-dimensional feature map based on a multi-head attention mechanism, performing image resolution recovery based on the different subspace information, and outputting a reconstructed image; physical compression is realized by using a mask, the data acquisition amount is greatly reduced, information recovery is performed through a deep learning reconstruction network, the image reconstruction efficiency and the perception quality are improved, and the method adapts to a multi-scene environment.
Need to check novelty before this filing date? Find Prior Art