High-resolution polarization images and multispectral data cubes of an object are difficult to capture simultaneously using traditional polarization, or multispectral imaging systems. Directly integrating band-pass filters into a polarization imaging system, or polarizers into a multispectral imaging system results in a low-light efficiency imaging system. In this paper, a high-efficiency multispectral polarization imaging system (HEMSPIS) has been proposed to acquire the spectral and polarization information of an object, simultaneously. By combining a notch filter array and a multi-camera imaging system, the polarization and multispectral information of an object are acquired simultaneously, and a compressive sensing-based method is employed to reconstruct the multispectral polarization data cube. The performance of our method has been demonstrated experimentally through observing polarization and multispectral images of polarization samples and natural senses. In comparison to the color polarization camera, the proposed imaging system provides a higher light efficiency and images with a higher signal-to-noise ratio.
Bandpass filter–based multispectral (MS) imaging systems have difficulty achieving high-quality MS imaging results while capturing high spatial resolution MS data cubes. This paper proposes a notch filter–based low-cost multicamera MS imaging system that acquires high-resolution MS images. By taking advantage of notch filters to block only specific bands of the spectrum, light from most of the spectrum is allowed to pass through, resulting in a high light efficiency imaging system. A compressive sensing approach is proposed to obtain images of high spatial and spectral resolution. Trained sparse dictionaries are used to perform the spectral and spatial data super-resolution of the acquired images. We simulated the effectiveness of our algorithm on a public dataset and verified the imaging performance of the prototype system by observing natural images. The experimental results show that the spatial resolution can be improved threefold on the laboratory target, the spectral resolution can be improved from 9 to 31 bands, and the average peak signal-to-noise ratio remains at 39. Our prototype imaging system can realize high spatial and spectral resolution imaging results.
Visible-light image and infrared image fusion technology can be used to obtain an image that has both the detailed texture of the visible-light image and the characteristics of infrared heat radiation. However, most existing fusion algorithms inevitably introduce noise from the source image during fusion. To solve this problem, an image fusion algorithm based on a multiscale transformation framework is proposed in this study. This algorithm first uses multiscale transformation theory to decompose the source images, and then uses a method based on the standard deviation of the local area as the fusion rule of the low-frequency components. In addition, a visual saliency detection method based on guided filtering is used to extract weight mapping of the high-frequency components, which are weighted and fused according to the distribution of the weight mapping. Finally, the components are inversely transformed to obtain a visible light–infrared fusion image. The experiments show that the proposed algorithm can effectively maintain the detailed texture of visible-light images, as well as infrared heat radiation information, and has the advantages of low noise, high contrast, and a better visual effect.
In the field of non-destructive testing (NDT), low-cost uncooled infrared (IR) sensors cannot generate high-speed thermography detection results with low levels of noise and high sensitivity. Hence, this paper proposes a compressive sensing-based postprocessing method for the results of uncooled IR sensors to obtain high-quality and super-resolution NDT imaging results. Super-resolution IR images are generated using low-resolution IR images and sparse dictionaries generated from randomly sampled raw patches of training images. Super-resolution IR images and pulsed phase thermography (PPT) are combined to improve the NDT results for carbon fiber-reinforced polymer (CFRP) specimens with artificial defects engraved on them. The results of the PPT experiments show that the compressive sensing-based super-resolution algorithm can be used to double the resolution of PPT phase images. The reconstructed PPT phase images further show that the proposed method can produce sharp defect edges while preserving the original texture, which will enable the use of uncooled IR sensors in a wider range of NDT applications.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.