Multi-Scale Structural-aware Exposure Correction for Endoscopic Imaging Axel Garc ıa-Vega1 Ricardo Espinosa24 Luis Ram ırez-Guzm an1 Thomas Bazin3 Luis Falc on-Morales1

2025-04-29 0 0 2.69MB 5 页 10玖币

侵权投诉

Multi-Scale Structural-aware Exposure Correction

for Endoscopic Imaging

Axel Garc´

ıa-Vega1, Ricardo Espinosa2,4, Luis Ram´

ırez-Guzm´

an1, Thomas Bazin3, Luis Falc´

on-Morales1,

Gilberto Ochoa-Ruiz1, Dominique Lamarque3and Christian Daul2,

Abstract—Endoscopy is the most widely used imaging tech-

nique for the diagnosis of cancerous lesions in hollow organs.

However, endoscopic images are often affected by illumination

artefacts: image parts may be over- or underexposed according

to the light source pose and the tissue orientation. These artifacts

have a strong negative impact on the performance of computer

vision or AI-based diagnosis tools. Although endoscopic image

enhancement methods are greatly required, little effort has been

devoted to over- and under-exposition enhancement in real-time.

This contribution presents an extension to the objective function

of LMSPEC, a method originally introduced to enhance images

from natural scenes. It is used here for the exposure correction in

endoscopic imaging and the preservation of structural informa-

tion. To the best of our knowledge, this contribution is the ﬁrst

one that addresses the enhancement of endoscopic images using

deep learning (DL) methods. Tested on the Endo4IE dataset, the

proposed implementation has yielded a signiﬁcant improvement

over LMSPEC reaching a SSIM increase of 4.40% and 4.21%

for over- and underexposed images, respectively.

Index Terms—image enhancement, endoscopy, exposure cor-

rection, Computer-Aided Diagnosis

I. INTRODUCTION

Endoscopy plays a central role in minimally invasive

surgery or for carrying out examinations in hollow organs,

such as the colon or the stomach. In recent years, computer

aided endoscopy has become an important area of research.

In particular, Computer Vision (CV) has the potential of

becoming an essential tool for assisting endoscopists in various

tasks [7], [14], [20].

However, a major hurdle that most of these CV methods

must face is related to the uncontrolled and highly changing

illumination conditions in endoscopic scenes. Figure 1 shows

two colonoscopic images in which strong illumination changes

are visible. Such uncontrolled lighting affects the robustness

of Computer-Aided Detection (CADe) and Diagnosis (CADx).

The performance of techniques for recovering extended sur-

faces of hollow organs (such as SLAM [8] or Structure for

Motion [15]) is also affected by uncontrolled lighting. These

strong photo-metric variations are due to non-optimal light

source poses, moist surfaces, and occlusions that lead to under-

or overexposed video-frames parts [13].

Identify applicable funding agency here. If none, delete this.

1School of Engineering and Sciences, Tecnologico de Monterrey, Mexico

2CRAN (UMR 7039, Universit´

e de Lorraine and CNRS), Nancy, France

3Hˆ

opital Ambroise Par´

e (AP-HP), Boulogne-Billancourt France

4Universidad Panamericana, Aguascalientes, Mexico

*Contacts: gilberto.ochoa@tec.mx, christian.daul@univ-lorraine.fr

Fig. 1: Strong illumination change example in almost con-

secutive frames of a colonoscopic image sequence. (a) This

image was acquired in appropriate lighting conditions. (b) Few

frames later, the image is overexposed in its lower left region

and underexposed in the remaining frame part.

Therefore, any improvements of endoscopic image content

quality could considerably boost the efﬁciency of CV- and AI-

based CAD tools. In this regard, various challenges have been

proposed in conferences to foster the development of algo-

rithms which can be bench-marked in terms of generalization

capabilities. One such challenge is the Endoscopic Artifact

Detection challenge (EAD, [3]), with includes various types of

endoscopic artefacts for developing novel image preprocessing

algorithms.

The results obtained by numerous methods in the EAD

challenge have shown that image enhancement (IE) algo-

rithms are of high interest for improving the robustness and

generalization capabilities of endoscopic image preprocessing

techniques. This contribution focuses on the exposure correc-

tion in white light endoscopy. It is noticeable that this issue

has only been partially addressed in the IE ﬁeld, as most

methods (see [1]) were dedicated to the correction of either

under- or over-exposed images, but did not deal with both

effects occurring concurrently. Contrary to images of natural

scenes, in endoscopy imaging it is common that both types of

non-optimal exposures simultaneously affect frames. Thus, a

preprocessing algorithm should be able to detect and correct

in real-time all types of inappropriate exposures.

Garc´

ıa-Vega et al. proposed a paired “normal-exposed” im-

age dataset [9], [10] to assess the ability of machine learning-

based methods to correct the effects of non-optimal lighting

conditions. However, the need for both accurate and real-

time IE techniques highlighted the shortcomings of most cur-

arXiv:2210.15033v1 [eess.IV] 26 Oct 2022

文档加载中……请稍候！
如果长时间未打开，您也可以点击刷新试试。

下载文档到电脑，查找使用更方便

10 玖币 0人已下载

立即下载

摘要：

Multi-ScaleStructural-awareExposureCorrectionforEndoscopicImagingAxelGarc´a-Vega1,RicardoEspinosa2;4,LuisRam´rez-Guzm´an1,ThomasBazin3,LuisFalc´on-Morales1,GilbertoOchoa-Ruiz1,DominiqueLamarque3andChristianDaul2,AbstractEndoscopyisthemostwidelyusedimagingtech-niqueforthediagnosisofcancerouslesion...

展开>> 收起<<

Multi-Scale Structural-aware Exposure Correction for Endoscopic Imaging Axel Garc ıa-Vega1 Ricardo Espinosa24 Luis Ram ırez-Guzm an1 Thomas Bazin3 Luis Falc on-Morales1.pdf

共5页,预览1页

还剩页未读，继续阅读

声明：本站为文档C2C交易模式，即用户上传的文档直接被用户下载，本站只是中间服务平台，本站所有文档下载所得的收益归上传人(含作者)所有。玖贝云文库仅提供信息存储空间，仅对用户上传内容的表现方式做保护处理，对上载内容本身不做任何修改或编辑。若文档所含内容侵犯了您的版权或隐私，请立即通知玖贝云文库，我们立即给予删除！

Multi-Scale Structural-aware Exposure Correction for Endoscopic Imaging Axel Garc ıa-Vega1 Ricardo Espinosa24 Luis Ram ırez-Guzm an1 Thomas Bazin3 Luis Falc on-Morales1

相关推荐

开通VIP享超值会员特权

作者详情

相关内容

热门标签

举报选择: