@article{MolloyMuellerDeeganetal.2024, author = {Molloy, Dara and M{\"u}ller, Patrick and Deegan, Brian and Mullins, Darragh and Horgan, Jonathan and Ward, Enda and Jones, Edward and Braun, Alexander and Glavin, Martin}, title = {Analysis of the Impact of Lens Blur on Safety-Critical Automotive Object Detection}, series = {IEEE Access}, volume = {12}, journal = {IEEE Access}, publisher = {IEEE}, issn = {2169-3536}, doi = {10.1109/ACCESS.2023.3348663}, url = {http://nbn-resolving.de/urn:nbn:de:hbz:due62-opus-42999}, pages = {3554 -- 3569}, year = {2024}, abstract = {Camera-based object detection is widely used in safety-critical applications such as advanced driver assistance systems (ADAS) and autonomous vehicle research. Road infrastructure has been designed for human vision, so computer vision, with RGB cameras, is a vital source of semantic information from the environment. Sensors, such as LIDAR and RADAR, are also often utilized for these applications; however, cameras provide a higher spatial resolution and color information. The spatial frequency response (SFR), or sharpness of a camera, utilized in object detection systems must be sufficient to allow a detection algorithm to localize objects in the environment over its lifetime reliably. This study explores the relationship between object detection performance and SFR. Six state-of-the-art object detection models are evaluated with varying levels of lens defocus. A novel raw image dataset is created and utilized, containing pedestrians and cars over a range of distances up to 100-m from the sensor. Object detection performance for each defocused dataset is analyzed over a range of distances to determine the minimum SFR necessary in each case. Results show that the relationship between object detection performance and lens blur is much more complex than previous studies have found due to lens field curvature, chromatic aberration, and astigmatisms. We have found that smaller objects are disproportionately impacted by lens blur, and different object detection models have differing levels of robustness to lens blur}, language = {en} } @article{BrummelMuellerBraun2022, author = {Brummel, Mattis and M{\"u}ller, Patrick and Braun, Alexander}, title = {Spatial precision and recall indices to assess the performance of instance segmentation algorithms}, series = {Electronic Imaging}, volume = {34}, journal = {Electronic Imaging}, number = {16}, publisher = {Society for Imaging Science and Technology}, issn = {2470-1173}, doi = {10.2352/EI.2022.34.16.AVM-101}, url = {http://nbn-resolving.de/urn:nbn:de:hbz:due62-opus-41552}, pages = {1 -- 6}, year = {2022}, language = {en} } @article{WohlersMuellerBraun2022, author = {Wohlers, Luis Constantin and M{\"u}ller, Patrick and Braun, Alexander}, title = {Original image noise reconstruction for spatially-varying filtered driving scenes}, series = {Electronic Imaging: Society for Imaging Science and Technology}, volume = {34}, journal = {Electronic Imaging: Society for Imaging Science and Technology}, number = {16}, publisher = {Society for Imaging Sciences and Technology}, issn = {2470-1173}, doi = {10.2352/EI.2022.34.16.AVM-214}, url = {http://nbn-resolving.de/urn:nbn:de:hbz:due62-opus-41479}, pages = {1 -- 7}, year = {2022}, language = {en} } @inproceedings{MuellerBraunKeuper2022, author = {M{\"u}ller, Patrick and Braun, Alexander and Keuper, Margret}, title = {Impact of realistic properties of the point spread function on classification tasks to reveal a possible distribution shift}, series = {NeurIPS 2022: Workshop on Distribution Shifts: Connecting Methods and Applications, December 3rd, 2022, New Orleans, USA}, booktitle = {NeurIPS 2022: Workshop on Distribution Shifts: Connecting Methods and Applications, December 3rd, 2022, New Orleans, USA}, address = {New Orleans}, year = {2022}, language = {en} } @inproceedings{MuellerBraun2022, author = {M{\"u}ller, Patrick and Braun, Alexander}, title = {Simulating optical properties to access novel metrological parameter ranges and the impact of different model approximations}, series = {2022 IEEE International Workshop on Metrology for Automotive (MetroAutomotive), 4-6 July 2022}, booktitle = {2022 IEEE International Workshop on Metrology for Automotive (MetroAutomotive), 4-6 July 2022}, publisher = {IEEE}, isbn = {978-1-6654-6689-9}, doi = {10.1109/MetroAutomotive54295.2022.9855079}, pages = {133 -- 138}, year = {2022}, language = {en} } @article{MuellerBraun2023, author = {M{\"u}ller, Patrick and Braun, Alexander}, title = {Local performance evaluation of AI-algorithms with the generalized spatial recall index}, series = {tm - Technisches Messen}, volume = {90}, journal = {tm - Technisches Messen}, number = {7-8}, publisher = {de Gruyter}, issn = {2196-7113}, doi = {10.1515/teme-2023-0013}, pages = {464 -- 477}, year = {2023}, language = {en} } @article{MuellerBraun2023, author = {M{\"u}ller, Patrick and Braun, Alexander}, title = {MTF as a performance indicator for AI algorithms?}, series = {Electronic Imaging: Society for Imaging Science and Technology}, volume = {35}, journal = {Electronic Imaging: Society for Imaging Science and Technology}, number = {16}, publisher = {Society for Imaging Sciences and Technology}, issn = {2470-1173}, doi = {10.2352/EI.2023.35.16.AVM-125}, pages = {1 -- 7}, year = {2023}, abstract = {Abstract The modulation-transfer function (MTF) is a fundamental optical metric to measure the optical quality of an imaging system. In the automotive industry it is used to qualify camera systems for ADAS/AD. Each modern ADAS/AD system includes evaluation algorithms for environment perception and decision making that are based on AI/ML methods and neural networks. The performance of these AI algorithms is measured by established metrics like Average Precision (AP) or precision-recall-curves. In this article we research the robustness of the link between the optical quality metric and the AI performance metric. A series of numerical experiments were performed with object detection and instance segmentation algorithms (cars, pedestrians) evaluated on image databases with varying optical quality. We demonstrate with these that for strong optical aberrations a distinct performance loss is apparent, but that for subtle optical quality differences - as might arise during production tolerances - this link does not exhibit a satisfactory correlation. This calls into question how reliable the current industry practice is where a produced camera is tested end-of-line (EOL) with the MTF, and fixed MTF thresholds are used to qualify the performance of the camera-under-test.}, language = {en} } @article{MuellerBrummelBraun2021, author = {M{\"u}ller, Patrick and Brummel, Mattis and Braun, Alexander}, title = {Spatial recall index for machine learning algorithms}, series = {London Imaging Meeting}, volume = {2021}, journal = {London Imaging Meeting}, number = {1}, publisher = {Society for Imaging Science and Technology}, isbn = {0-89208-346-6}, doi = {10.2352/issn.2694-118X.2021.LIM-58}, url = {http://nbn-resolving.de/urn:nbn:de:hbz:due62-opus-35657}, pages = {58 -- 62}, year = {2021}, abstract = {We present a novel metric Spatial Recall Index to assess the performance of machine-learning (ML) algorithms for automotive applications, focusing on where in the image which performance occurs. Typical metrics like intersection-over-union (IoU), precisionrecallcurves or average precision (AP) quantify the performance over a whole database of images, neglecting spatial performance variations. But as the optics of camera systems are spatially variable over the field of view, the performance of ML-based algorithms is also a function of space, which we show in simulation: A realistic objective lens based on a Cooke-triplet that exhibits typical optical aberrations like astigmatism and chromatic aberration, all variable over field, is modeled. The model is then applied to a subset of the BDD100k dataset with spatially-varying kernels. We then quantify local changes in the performance of the pre-trained Mask R-CNN algorithm. Our examples demonstrate the spatial dependence of the performance of ML-based algorithms from the optical quality over field, highlighting the need to take the spatial dimension into account when training ML-based algorithms, especially when looking forward to autonomous driving applications.}, language = {en} } @article{KrebsMuellerBraun2021, author = {Krebs, Christian and M{\"u}ller, Patrick and Braun, Alexander}, title = {Impact of Windshield Optical Aberrations on Visual Range Camera Based Classification Tasks Performed by CNNs}, series = {London Imaging Meeting}, volume = {2021}, journal = {London Imaging Meeting}, number = {1}, publisher = {Ingenta connect}, isbn = {0-89208-346-6}, issn = {2694-118X}, doi = {10.2352/issn.2694-118X.2021.LIM-83}, pages = {83 -- 87}, year = {2021}, language = {de} } @inproceedings{MuellerLehmannBraun2019, author = {M{\"u}ller, Patrick and Lehmann, Matthias and Braun, Alexander}, title = {Optical quality metrics for image restoration}, series = {Digital Optical Technologies 2019}, volume = {Proceedings, Vol. 11062}, booktitle = {Digital Optical Technologies 2019}, editor = {Kress, Bernard C. and Schelkens, Peter}, address = {Munich}, organization = {International Society for Optics and Photonics}, doi = {10.1117/12.2528100}, pages = {1106214}, year = {2019}, abstract = {Image restoration is a process used to remove blur (from different sources like object motion or aberrations) from images by either non-blind or blind-deconvolution. The metrics commonly used to quantify the restoration process are peak signal-to-noise ratio (PSNR) and structural similarity index measure (SSIM). Often only a small sample of test images are used (like Lena or the camera guy). In optical design research PSNR and SSIM are not normally used, here image quality metrics based on linear system theory (e.g. modulation transfer function, MTF) are used to quantify optical errors like spherical or chromatic aberration. In this article we investigate how different image restoration algorithms can be quantified by applying image quality metrics. We start with synthetic image data that is used in camera test stands (e.g. Siemens star etc.), apply two different spatially variant degradation algorithms, and restore the original image by a direct method (Wiener filtering within sub-images), and by an iterative method (alternating direction method of multipliers, ADMM). Afterwards we compare the quality metrics (like MTF curves) for the original, the degraded and the restored image. As a first result we show that restoration algorithms sometimes fail in dealing with non-natural scenes, e.g. slanted-edge targets. Further, these first results indicate a correlation between degradation and restoration, i.e. the restoration algorithms are not capable of removing the optically relevant errors introduced by the degradation, a fact neither visible nor available from the PSNR values. We discuss the relevance in the context of the automotive industry, where image restoration may yield distinct advantages for camera-based applications, but testing methods rely on the used image quality metrics.}, language = {en} }