Image Domain Research Articles

Frontiers and developments of data augmentation for image: From unlearnable to learnable

Data augmentation is a crucial technique for expanding training datasets, effectively alleviating the overfitting issue that arises from limited training data in deep learning models. This paper takes a fresh perspective and offers a scholarly exploration of image data augmentation, following a logical progression from unlearnable to learnable methods. The paper begins by providing a brief overview of the developmental history of data augmentation. It categorizes data augmentation techniques into unlearnable and learnable based on their “variation” strategies. Furthermore, the paper outlines the fundamental properties of data augmentation, including expansiveness, fidelity, generalizability, and self-adaptability. Subsequently, focusing on unlearnable and learnable data augmentation techniques, the paper further divides them into single-sample and multi-sample, global and local, image domain, and feature domain, categorically reviewing the basic principles and effects of various data augmentation methods based on the differences in the sources, scopes, and content of “variation” attributes. Ultimately, the comparative analysis of diverse data augmentation methodologies in specific tasks is conducted alongside a synthesis and projection of future research directions. By comprehensively analyzing diverse image data augmentation methods from a fresh perspective, this review reveals the intrinsic disparities between unlearnable and learnable data augmentation techniques. It paves the way for scholars to embark on innovative paths in data augmentation.

Advancing lung ultrasound imaging: advanced pre-processing and neoteric segmentation for improved accuracy

Abstract Within the medical imaging domain, the segmentation of Lung Ultrasound (LUS) images is crucial for accurately defining regions of interest, particularly for lung health assessments. various techniques such as threshold-based, region-based, and edge-based methods have been employed to address this challenge, their accuracy and efficiency are often questioned. This highlights the need for an innovative approach. With this context in mind, this research introduces a novel preprocessing step using Optimized Gaussian Histogram Equalization (GHE) followed by a Neoteric Segmentation (NS) approach. This combination enhances the capability of segmenting specific regions within LUS images, regardless of the overarching medical scenario. The primary focus of this method is to understand, through a thorough evaluation, how neoteric segmentation compares with more conventional methods. By conducting an in-depth comparative analysis, valuable insights can be unearthed. The results demonstrate that the optimized GHE technique significantly enhances contrast and preserves visual quality, while the NS approach shows superior performance in accurately segmenting the lung regions. Initial findings suggest that the NS method, achieving an impressive 95% accuracy, holds significant promise and could influence the trajectory of healthcare practices. This research aims to advance the field of medical image analysis by highlighting the potential benefits of this cutting-edge segmentation technique. These advancements could positively impact healthcare outcomes, especially in contexts that leverage lung ultrasound imaging.

Image-domain least-squares imaging in an elastic vertically transversely isotropic medium: Multiparameter point-spread function deconvolution

By solving a linear inverse problem, least-squares migration (LSM) can provide higher-precision images of complex subsurface structures than traditional adjoint-operator-based migration. LSM can be conducted in the data domain and the image domain (IDLSM). IDLSM in acoustic media has been extensively studied, commonly using single-parameter point-spread function (PSF) deconvolution. For high-accuracy imaging of complex media, we develop an image-domain elastic least-squares imaging method using multiparameter PSF deconvolution for vertically transversely isotropic (VTI) media. The multiparameter PSFs, including P-P, P-S, S-P, and S-S components, are calculated on a sparse grid using elastic VTI wave equation Born modeling and migration. The traditional migration result is then decomposed into individual local results, each corresponding to the size of a single PSF, through unit partitioning. Finally, a set of filters is derived from four elastic VTI PSF components to approximate the Hessian inverse, which are subsequently assembled to reconstruct an entire deconvolved image. The numerical experiments confirm that our method can significantly decrease multiparameter crosstalk artifacts and improve spatial resolution and amplitude fidelity.

FPL+: Filtered Pseudo Label-Based Unsupervised Cross-Modality Adaptation for 3D Medical Image Segmentation.

Adapting a medical image segmentation model to a new domain is important for improving its cross-domain transferability, and due to the expensive annotation process, Unsupervised Domain Adaptation (UDA) is appealing where only unlabeled images are needed for the adaptation. Existing UDA methods are mainly based on image or feature alignment with adversarial training for regularization, and they are limited by insufficient supervision in the target domain. In this paper, we propose an enhanced Filtered Pseudo Label (FPL+)-based UDA method for 3D medical image segmentation. It first uses cross-domain data augmentation to translate labeled images in the source domain to a dual-domain training set consisting of a pseudo source-domain set and a pseudo target-domain set. To leverage the dual-domain augmented images to train a pseudo label generator, domain-specific batch normalization layers are used to deal with the domain shift while learning the domain-invariant structure features, generating high-quality pseudo labels for target-domain images. We then combine labeled source-domain images and target-domain images with pseudo labels to train a final segmentor, where image-level weighting based on uncertainty estimation and pixel-level weighting based on dual-domain consensus are proposed to mitigate the adverse effect of noisy pseudo labels. Experiments on three public multi-modal datasets for Vestibular Schwannoma, brain tumor and whole heart segmentation show that our method surpassed ten state-of-the-art UDA methods, and it even achieved better results than fully supervised learning in the target domain in some cases.

Telescope imaging beyond the Rayleigh limit in extremely low SNR

Abstract The Rayleigh limit and low Signal-to-Noise Ratio (SNR) scenarios pose significant limitations to optical imaging systems used in remote sensing, infrared thermal imaging, and space domain awareness. In this study, we introduce a Stochastic Sub-Rayleigh Imaging (SSRI) algorithm to localize point objects and estimate their positions, brightnesses, and number in low SNR conditions, even below the Rayleigh limit. Our algorithm adopts a maximum likelihood approach and exploits the Poisson distribution of incoming photons to overcome the Rayleigh limit in low SNR conditions. In our experimental validation, which closely mirrors practical scenarios, we focus on conditions with closely spaced sources within the sub-Rayleigh limit (0.49-1.00R) and weak signals (SNR less than 1.2). We use the Jaccard index and Jaccard efficiency as a figure of merit to quantify imaging performance in the sub-Rayleigh region. Our approach consistently outperforms established algorithms such as Richardson-Lucy and CLEAN by 4X in the low SNR, sub-Rayleigh regime. Our SSRI algorithm allows existing telescope-based optical/infrared imaging systems to overcome the extreme limit of sub-Rayleigh, low SNR source distributions, potentially impacting a wide range of fields, including passive thermal imaging, remote sensing, and space domain awareness.

MTC-CSNet: Marrying Transformer and Convolution for Image Compressed Sensing.

Image compressed sensing (ICS) has been extensively applied in various imaging domains due to its capability to sample and reconstruct images at subNyquist sampling rates. The current predominant approaches in ICS, specifically pure convolutional networks (ConvNets)-based ICS methods, have demonstrated their effectiveness in capturing local features for image recovery. Simultaneously, the Transformer architecture has gained significant attention due to its capability to model global correlations among image features. Motivated by these insights, we propose a novel hybrid network for ICS, named MTC-CSNet, which effectively combines the strengths of both ConvNets and Transformer architectures in capturing local and global image features to achieve high-quality image recovery. Particularly, MTC-CSNet is a dual-path framework that consists of a ConvNets-based recovery branch and a Transformer-based recovery branch. Along the ConvNets-based recovery branch, we design a lightweight scheme to capture the local features in natural images. Meanwhile, we implement a Transformer-based recovery branch to iteratively model the global dependencies among image patches. Ultimately, the ConvNets-based and Transformer-based recovery branches collaborate through a bridging unit, facilitating the adaptive transmission and fusion of informative features for image reconstruction. Extensive experimental results demonstrate that our proposed MTC-CSNet surpasses the state-of-the-art methods on various public datasets. The code and models are publicly available at MTC-CSNet.

DuDoCFNet: Dual-Domain Coarse-to-Fine Progressive Network for Simultaneous Denoising, Limited-View Reconstruction, and Attenuation Correction of Cardiac SPECT.

Single-Photon Emission Computed Tomography (SPECT) is widely applied for the diagnosis of coronary artery diseases. Low-dose (LD) SPECT aims to minimize radiation exposure but leads to increased image noise. Limited-view (LV) SPECT, such as the latest GE MyoSPECT ES system, enables accelerated scanning and reduces hardware expenses but degrades reconstruction accuracy. Additionally, Computed Tomography (CT) is commonly used to derive attenuation maps ( μ -maps) for attenuation correction (AC) of cardiac SPECT, but it will introduce additional radiation exposure and SPECT-CT misalignments. Although various methods have been developed to solely focus on LD denoising, LV reconstruction, or CT-free AC in SPECT, the solution for simultaneously addressing these tasks remains challenging and under-explored. Furthermore, it is essential to explore the potential of fusing cross-domain and cross-modality information across these interrelated tasks to further enhance the accuracy of each task. Thus, we propose a Dual-Domain Coarse-to-Fine Progressive Network (DuDoCFNet), a multi-task learning method for simultaneous LD denoising, LV reconstruction, and CT-free μ -map generation of cardiac SPECT. Paired dual-domain networks in DuDoCFNet are cascaded using a multi-layer fusion mechanism for cross-domain and cross-modality feature fusion. Two-stage progressive learning strategies are applied in both projection and image domains to achieve coarse-to-fine estimations of SPECT projections and CT-derived μ -maps. Our experiments demonstrate DuDoCFNet's superior accuracy in estimating projections, generating μ -maps, and AC reconstructions compared to existing single- or multi-task learning methods, under various iterations and LD levels. The source code of this work is available at https://github.com/XiongchaoChen/DuDoCFNet-MultiTask.

Fekete-Szegö and Hankel inequalities related to the sine function

ABSTRACT The Fekete-Szegö inequality is one of the inequalities for the coefficients which associated with the famous Bieberbach conjecture. Other issues associated with this inequality are determining the Hankel determinant denoted as Hd inequalities which are involved in the study of the singularities and integral coefficients in the Taylor series representations. In this investigation, we discuss such inequalities for certain mappings g for which ( g ′ ( z ) ) α [ 2 z g ′ ( z ) g ( z ) − g ( − z ) ] ( 1 − α ) ≺ 1 + sin ⁡ z , ( 0 ≤ α ≤ 1 ) lies in the image domain starlike about 1 and symmetric about real axis. Particularly, we obtain certain inequality for functions involving Poisson distribution. Furthermore, we also find the second Hankel determinant and other inequalities related to this newly defined class.

Impact of scattering phase function and polarization on the accuracy of diffuse and sub-diffuse spatial frequency domain imaging.

Although spatial frequency domain imaging (SFDI) has been well characterized under diffuse optical conditions, tissue measurements made outside the diffuse regime can provide new diagnostic information. Before such measurements can become clinically relevant, however, the behavior of sub-diffuse SFDI and its effect on the accuracy of derived tissue parameters must be assessed. We aim to characterize the impact that both the assumed scattering phase function (SPF) and the polarization state of the illumination light source have on the accuracy of SFDI-derived optical properties when operating under diffuse or sub-diffuse conditions, respectively. Through the use of a set of well-characterized optical phantoms, SFDI accuracy was assessed at four wavelengths (395, 545, 625, and 850nm) and two different spatial frequencies (0.3 and ), which provided a broad range of diffuse and sub-diffuse conditions, using three different SPFs. To determine the effects of polarization, the SFDI accuracy was assessed using both unpolarized and cross-polarized illumination. It was found that the assumed SPF has a direct and significant impact on the accuracy of the SFDI-derived optical properties, with the best choice of SPF being dictated by the polarization state. As unpolarized SFDI retains the sub-diffuse portion of the signal, optical properties were found to be more accurate when using the full SPF that includes forward and backscattering components. By contrast, cross-polarized SFDI yielded accurate optical properties when using a forward-scattering SPF, matching the behavior of cross-polarization to attenuate the immediate backscattering of sub-diffuse reflectance. Using the correct pairings of SPF and polarization enabled using a reflectance standard, instead of a more subjective phantom, as the reference measurement. These results provide the foundation for a more thorough understanding of SFDI and enable new applications of this technology in which sub-diffuse conditions dominate (e.g., ) or high spatial frequencies are required.

The Resident Space Object Detection Method Based on the Connection between the Fourier Domain Image of the Video Data Difference Frame and the Orbital Velocity Projection

The Resident Space Object Detection Method Based on the Connection between the Fourier Domain Image of the Video Data Difference Frame and the Orbital Velocity Projection

Optical frequency domain imaging-guided versus intravascular ultrasound-guided percutaneous coronary intervention for acute coronary syndromes: the OPINION ACS randomised trial.

The clinical benefits of optical frequency domain imaging (OFDI)-guided percutaneous coronary intervention (PCI) for acute coronary syndrome (ACS) remain unclear. We sought to compare intravascular ultrasound (IVUS)- and OFDI-guided PCI in patients with ACS. OPINION ACS is a multicentre, prospective, randomised, non-inferiority trial that compared OFDI-guided PCI with IVUS-guided PCI using current-generation drug-eluting stents in ACS patients (n=158). The primary endpoint was in-stent minimum lumen area (MLA), assessed using 8-month follow-up OFDI. Patients presented with ST-segment elevation myocardial infarction (55%), non-ST-segment elevation myocardial infarction (29%), or unstable angina pectoris (16%). PCI procedural success was achieved in all patients, with comparably low periprocedural complications rates in both groups. Immediately after PCI, the minimum stent area (p=0.096) tended to be smaller for OFDI versus IVUS guidance. Proximal stent edge dissection (p=0.012) and irregular protrusion (p=0.03) were significantly less frequent in OFDI-guided procedures than in IVUS-guided procedures. Post-PCI coronary flow, assessed using corrected Thrombolysis in Myocardial Infarction frame counts, was significantly better in the OFDI-guided group than in the IVUS-guided group (p<0.001). The least squares mean (95% confidence interval [CI]) in-stent MLA at 8 months was 4.91 (95% CI: 4.53-5.30) mm2 and 4.76 (95% CI: 4.35-5.17) mm2 in the OFDI- and IVUS-guided groups, respectively, demonstrating the non-inferiority of OFDI guidance (pnon-inferiority<0.001). The average neointima area tended to be smaller in the OFDI-guided group. The frequency of major adverse cardiac events was similar. Among ACS patients, OFDI-guided PCI and IVUS-guided PCI were equally safe and feasible, with comparable in-stent MLA at 8 months. OFDI guidance may be a potential option in ACS patients. This study was registered in the Japan Registry of Clinical Trials (jrct.niph.go.jp: jRCTs052190093).

Deep Interactive Segmentation of Medical Images: A Systematic Review and Taxonomy.

Interactive segmentation is a crucial research area in medical image analysis aiming to boost the efficiency of costly annotations by incorporating human feedback. This feedback takes the form of clicks, scribbles, or masks and allows for iterative refinement of the model output so as to efficiently guide the system towards the desired behavior. In recent years, deep learning-based approaches have propelled results to a new level causing a rapid growth in the field with 121 methods proposed in the medical imaging domain alone. In this review, we provide a structured overview of this emerging field featuring a comprehensive taxonomy, a systematic review of existing methods, and an in-depth analysis of current practices. Based on these contributions, we discuss the challenges and opportunities in the field. For instance, we find that there is a severe lack of comparison across methods which needs to be tackled by standardized baselines and benchmarks.

Stokes Phenomena in Lensing

As lensing of coherent astrophysical sources, e.g., pulsars, fast radio bursts, and gravitational waves, becomes observationally relevant, the mathematical framework of Picard–Lefschetz theory has recently been introduced to fully account for wave optics effects. Accordingly, the concept of lensing images has been generalized to include complex solutions of the lens equation referred to as “imaginary images,” and more radically, to the Lefschetz thimbles, which are a sum of the steepest descent contours connecting the real and imaginary images in the complex domain. In this wave-optics-based theoretical framework of lensing, we study the “Stokes phenomena” as the change of the topology of the Lefschetz thimbles. Similar to the well-known caustics at which the number of geometric images changes abruptly, the corresponding Stokes lines are the boundaries in the parameter space where the number of effective imaginary images changes. We map the Stokes lines for a few lens models. The resulting Stokes line-caustics network represents a unique feature of the lens models. The observable signature of the Stokes phenomena is the change of interference behavior, in particular the onset of frequency oscillation for some Stokes lines. We also demonstrate high-order Stokes phenomena where the system has a continuous number of effective images but with an abrupt change in the way they are connected to each other by the Lefschetz thimbles. Their full characterization calls for an analogy of the catastrophe theory for caustics.

Research on fractional wavelet transform combined with parameter adaptive PCNN for infrared and visible image fusion algorithm

Infrared and visible image fusion holds significant value across various fields due to its ability to provide complementary information. However, existing fusion algorithms often suffer from susceptibility to external physical conditions of source images, leading to inconsistent fusion quality and limited adaptive capability. In this paper, we propose a fractional wavelet transform fusion algorithm for infrared and visible images. This algorithm uses fractional wavelet domain image decomposition to more accurately locate the image structures of different scales. Initially, we perform image decomposition using a non-subsampled shearlet transform (NSST) with multi-scale and multi-directional decomposition. Subsequently, to effectively fuse detailed information, we employ discrete fractional wavelet transform (DFRWT) to decompose low-frequency subbands while selecting an appropriate p-value. High-frequency subbands are fused using a parameterized adaptive pulse-coupled neural network (PCNN) model. To validate the superiority of our algorithm, we conduct visual and quantitative comparative analyses against several existing algorithms. The results confirm that our proposed algorithm exhibits robustness against external physical conditions and surpasses these existing algorithms, making it highly applicable for infrared–visible fusion tasks.

Tunable Privacy Risk Evaluation of Generative Adversarial Networks.

Generative machine learning models such as Generative Adversarial Networks (GANs) have been shown to be especially successful in generating realistic synthetic data in image and tabular domains. However, it has been shown that such generative models, as well as the generated synthetic data, can reveal information contained in their privacy-sensitive training data, and therefore must be carefully evaluated before being used. The gold standard method through which such privacy leakage can be estimated is simulating membership inference attacks (MIAs), in which an attacker attempts to learn whether a given sample was part of the training data of a generative model. The state-of-the art MIAs against generative models, however, rely on strong assumptions (knowledge of the exact training dataset size), or require a lot of computational power (to retrain many "surrogate" generative models), which make them hard to use in practice. In this work, we propose a technique for evaluating privacy risks in GANs which exploits the outputs of the discriminator part of the standard GAN architecture. We evaluate our attacks in terms of performance in two synthetic image generation applications in radiology and ophthalmology, showing that our technique provides a more complete picture of the threats by performing worst-case privacy risk estimation and by identifying attacks with higher precision than the prior work.

Detection and localization of the highly active FRB 20240114A with MeerKAT

ABSTRACT We report observations of the highly active FRB 20240114A with MeerKAT using the ultra-high frequency (UHF; $544\!-\!1088$ MHz) and L band ($856\!-\!1712$ MHz) receivers. A total of 62 bursts were detected in coherent tied-array beams using the MeerTRAP real-time transient detection pipeline. We measure a structure-optimizing dispersion measure of $527.65\pm 0.01\, \text{pc}\, \text{cm}^{-3}$ using the brightest burst in the sample. We find the bursts of FRB 20240114A are generally detected in part of the broad-band of MeerKAT, $\sim 40~{{\ \rm per\ cent}}$ in the UHF and $\sim 30~{{\ \rm per\ cent}}$ in the L band, indicating the band limited nature. We analyse the fluence distribution of the 44 bursts detected at UHF, constraining the fluence completeness limit to $\sim 1$ Jy ms, above which the cumulative burst rate follows a power law $R (\gt F)\propto (F/1\, \text{Jy}\, \text{ms})^\gamma$ with $\gamma =-1.8\pm 0.2$. Using channelized telescope data captured in our transient buffer we localize FRB 20240114A in the image domain to RA = 21h27m39.86s, Dec. = +04d19m45.01s with an uncertainty of 1.4 arcsec. This localization allows us to confidently identify the host galaxy of FRB 20240114A. Also using the transient buffer data, we perform a polarimetric study and demonstrate that most of the bursts have $\sim 100~{{\ \rm per\ cent}}$ linear polarization fractions and up to $\sim 20~{{\ \rm per\ cent}}$ circular polarization fractions. Finally, we predict the flux density of a potential persistent radio source (PRS) associated with FRB 20240114A is $\backsimeq [0.6\!-\!60]\, \mu \text{Jy}$ based on the simple relation between the luminosity of the PRS and the rotation measure arising from the FRB local environment.

MA-SAM: Modality-agnostic SAM adaptation for 3D medical image segmentation

The Segment Anything Model (SAM), a foundation model for general image segmentation, has demonstrated impressive zero-shot performance across numerous natural image segmentation tasks. However, SAM’s performance significantly declines when applied to medical images, primarily due to the substantial disparity between natural and medical image domains. To effectively adapt SAM to medical images, it is important to incorporate critical third-dimensional information, i.e., volumetric or temporal knowledge, during fine-tuning. Simultaneously, we aim to harness SAM’s pre-trained weights within its original 2D backbone to the fullest extent. In this paper, we introduce a modality-agnostic SAM adaptation framework, named as MA-SAM, that is applicable to various volumetric and video medical data. Our method roots in the parameter-efficient fine-tuning strategy to update only a small portion of weight increments while preserving the majority of SAM’s pre-trained weights. By injecting a series of 3D adapters into the transformer blocks of the image encoder, our method enables the pre-trained 2D backbone to extract third-dimensional information from input data. We comprehensively evaluate our method on five medical image segmentation tasks, by using 11 public datasets across CT, MRI, and surgical video data. Remarkably, without using any prompt, our method consistently outperforms various state-of-the-art 3D approaches, surpassing nnU-Net by 0.9%, 2.6%, and 9.9% in Dice for CT multi-organ segmentation, MRI prostate segmentation, and surgical scene segmentation respectively. Our model also demonstrates strong generalization, and excels in challenging tumor segmentation when prompts are used. Our code is available at: https://github.com/cchen-cc/MA-SAM.

Interdisciplinary provider visits attenuate relationship between patient concerns and distress in older adults with cancer.

This study examined the relationship between multidimensional patient concerns and anxiety and depression in a national sample of older adults with cancer (OACs ≥ 65years) and the buffering effect of visiting providers across disciplines (e.g., oncology, allied health, primary care, mental health) on these relationships. Participants completed a cross-sectional survey through the Cancer Support Community's Cancer Experience Registry (CER), an online community-based research initiative. Eligible participants were 65years and older and diagnosed with cancer in the past five years. Participants completed self-report measures of (1) the severity of their concerns across multiple domains, (2) anxiety and depression, and (3) whether they received care for "symptoms and side effects" from various providers. The sample consisted of 277 OACs; 45% endorsed elevated anxiety and 31% endorsed elevated depression. The most severe concerns were in the domains of body image and healthy lifestyle and symptom burden and impact. More severe concerns were associated with higher levels of anxiety and depression. The relationship between concern severity and distress was weaker in OACs who saw a palliative care, mental health, physical or occupational therapy provider, pharmacist, or primary care provider relative to OACs who did not. A visit with an oncology provider did not moderate most relationships between concerns and distress. The relationship between OACs' concerns and distress was attenuated by treatment with a specialty provider. Interdisciplinary team care may be a vital component of comprehensive patient-centered care for OACs.

A REVIEW ON HYPERSPECTRAL IMAGING AND ITS APPLICATIONS IN WOOD DOMAIN

A REVIEW ON HYPERSPECTRAL IMAGING AND ITS APPLICATIONS IN WOOD DOMAIN

Soybean seed pest damage detection method based on spatial frequency domain imaging combined with RL-SVM

Soybean seeds are susceptible to damage from the Riptortus pedestris, which is a significant factor affecting the quality of soybean seeds. Currently, manual screening methods for soybean seeds are limited to visual inspection, making it difficult to identify seeds that are phenotypically defect-free but have been punctured by stink bugs on the sub-surface. To facilitate the convenient and efficient identification of healthy soybean seeds, this paper proposes a soybean seed pest detection method based on spatial frequency domain imaging combined with RL-SVM. Firstly, soybean optical data is obtained using single integration sphere technique, and the vigor index of soybean seeds is obtained through germination experiments. Then, based on the above two data items using feature extraction algorithms (the successive projections algorithm and the competitive adaptive reweighted sampling algorithm), the characteristic wavelengths of soybeans are identified. Subsequently, the spatial frequency domain imaging technique is used to obtain the sub-surface images of soybean seeds in a forward manner, and the optical coefficients such as the reduced scattering coefficient μ′s\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$${{\\mu }{\\prime}}_{s}$$\\end{document} and absorption coefficient μa\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$${\\mu }_{a}$$\\end{document} of soybean seeds are inverted. Finally, RL-MLR, RL-GRNN, and RL-SVM prediction models are established based on the ratio of the area of insect-damaged sub-surface to the entire seed, soybean varieties, and μa\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$${\\mu }_{a}$$\\end{document} at three wavelengths (502 nm, 813 nm, and 712 nm) for predicting and identifying soybean the stinging and sucking pest damage levels of soybean seeds. The experimental results show that the spatial frequency domain imaging technique yields small errors in the optical coefficients of soybean seeds, with errors of less than 15% for μa\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$${\\mu }_{a}$$\\end{document} and less than 10% for μ′s\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$${{\\mu }{\\prime}}_{s}$$\\end{document}. After parameter adjustment through reinforcement learning, the Macro-Recall metrics of each model have improved by 10%-15%, and the RL-SVM model achieves a high Macro-Recall value of 0.9635 for classifying the pest damage levels of soybean seeds.