Deep Photometric Stereo Networks for Determining Surface Normal and Reflectances

Santo, Hiroaki; Samejima, Masaki; Sugano, Yusuke; Shi, Boxin; Matsushita, Yasuyuki

doi:10.1109/tpami.2020.3005219

Cited by 15 publications

(6 citation statements)

References 55 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, as a result of three-dimensional image processing of the artery using a photometric method, it became difficult to observe three-dimensional imaging with green light. This is because the photometric method [ 29 , 30 ] is a technique to synthesize three-dimensional images from planar images captured from different directions. As the wavelength of the photometric method is sufficient accurate for small veins on the surface layer, it is difficult to create three-dimensional images owing to the longer wavelength compared to that of blue light.…”

Section: Resultsmentioning

confidence: 99%

Quantification and Visualization of Reliable Hemodynamics Evaluation Based on Non-Contact Arteriovenous Fistula Measurement

Iwai

Shimazaki

Kawakubo

et al. 2022

Sensors

View full text Add to dashboard Cite

The condition of arteriovenous fistula (AVF) blood flow is typically checked by using auscultation; however, auscultation should require a qualitative judgment dependent on the skills of doctors, and further attention to contact infection is required. For these reasons, this study developed a non-contact and non-invasive medical device to measure the pulse wave of AVFs by applying optical imaging technology. As a first step toward realization of the quantification judgment based on non-contact AVF measurement, we experimentally validated the developed system, whereby the hemodynamics of 168 subjects were visually and quantitatively evaluated based on clinical tests. Based on the evaluation results, the fundamental statistical characteristics of the non-contact measurement, including the average and median values, and distribution of measured signal-to-noise power ratio, were demonstrated. The clinical test results contributed to the future construction of quantified criteria for the AVF condition with the non-contact measurement.

show abstract

Section: Resultsmentioning

confidence: 99%

Quantification and Visualization of Reliable Hemodynamics Evaluation Based on Non-Contact Arteriovenous Fistula Measurement

Iwai

Shimazaki

Kawakubo

et al. 2022

Sensors

View full text Add to dashboard Cite

show abstract

“…sR sG sB (16) where most entries in f i,j s are near-zero since images are often near specular-free. In such a case, a non-zero residual occurs if f i,j s contains non-negligible entries.…”

Section: Compute Residual Vectormentioning

confidence: 99%

“…To widen the applicability, recent years have seen non-parametric BRDFs based on machine learning. Santo et al [15,16] used a deep neural network for the first time whereas Taniai and Maehara [17] estimated surface normals and BRDFs by unsupervised learning. Ikehata [18] estimated surface normals more straightforward by deriving the so-called observation maps and using convolutional neural networks.…”

Section: Introductionmentioning

confidence: 99%

DRM-Based Colour Photometric Stereo Using Diffuse-Specular Separation for Non-Lambertian Surfaces

Furukawa

2022

J. Imaging

View full text Add to dashboard Cite

This paper presents a photometric stereo (PS) method based on the dichromatic reflectance model (DRM) using colour images. The proposed method estimates surface orientations for surfaces with non-Lambertian reflectance using diffuse-specular separation and contains two steps. The first step, referred to as diffuse-specular separation, initialises surface orientations in a specular invariant colour subspace and further separates the diffuse and specular components in the RGB space. In the second step, the surface orientations are refined by first initialising specular parameters via solving a log-linear regression problem owing to the separation and then fitting the DRM using Levenburg-Marquardt algorithm. Since reliable information from diffuse reflection free from specularities is adopted in the initialisations, the proposed method is robust and feasible with less observations. At pixels where dense non-Lambertian reflectances appear, signals from specularities are exploited to refine the surface orientations and the additionally acquired specular parameters are potentially valuable for more applications, such as digital relighting. The effectiveness of the newly proposed surface normal refinement step was evaluated and the accuracy in estimating surface orientations was enhanced around 30% on average by including this step. The proposed method was also proven effective in an experiment using synthetic input images comprised of twenty-four different reflectances of dielectric materals. A comparison with nine other PS methods on five representative datasets further prove the validity of the proposed method.

show abstract

“…The learning based approaches comprise several methods which can be further classified into two classes. The first class attempts to solve the problem of uncalibrated PS using only the collected images [10][11][12][13]. The second class of learning based methods utilizes other available information such as light positions or surface characteristics as inputs additionally [14][15][16].…”

Section: Introductionmentioning

confidence: 99%

Photometric Stereo of Surfaces with Heterogeneous Reflection using Adaptive Diffuse-Specular Segregation

Singh

Furukawa

2023

Preprint

View full text Add to dashboard Cite

Photometric Stereo (PS) techniques have been developed accounting for several material and surface characteristics. While various methods have achieved high accuracy, this has come at the cost of large datasets or intricate reflectance models. Statistical PS methods attempt to estimate the normals robustly, however, they are limited to sparsely specular surfaces. In this paper, we introduce a novel Adaptive Principal Component Analysis (APCA) based PS approach that can estimate the normals of surfaces with heterogeneous reflectance properties with a minimal number of images. The proposed approach adaptively estimates the dominant reflectance characteristic at each individual pixel and solves an optimization problem to estimate the surface normals. The effectiveness of the method is validated using both simulated and real-world data comparing it against representative past methods. The ability of the proposed method to handle heterogenous reflectance was verified. The proposed method shows improvement for surfaces with dominant heterogeneous reflectance. The repeatability and applicability of the proposed method are also verified using data collected by the authors.

show abstract

Deep Photometric Stereo Networks for Determining Surface Normal and Reflectances

Cited by 15 publications

References 55 publications

Quantification and Visualization of Reliable Hemodynamics Evaluation Based on Non-Contact Arteriovenous Fistula Measurement

Quantification and Visualization of Reliable Hemodynamics Evaluation Based on Non-Contact Arteriovenous Fistula Measurement

DRM-Based Colour Photometric Stereo Using Diffuse-Specular Separation for Non-Lambertian Surfaces

Photometric Stereo of Surfaces with Heterogeneous Reflection using Adaptive Diffuse-Specular Segregation

Contact Info

Product

Resources

About