Image Points Research Articles

Semi-supervised latent diffusion model for Biliary Atresia class-imbalanced image recognition

Biliary Atresia (BA) is a congenital condition affecting the bile ducts in newborn babies worldwide, typically diagnosed by the color of infant stools in clinical practice. Due to the limited availability of pediatricians and the lack of effective diagnostic tools for parents, computer vision is devised as a non-invasive and convenient means for early detection of BA. However, the limited recognition accuracy of BA and the challenge of imbalanced classes in real-world scenarios restrict the generalizability of existing methods, posing a challenge to their decision-making performance. To address those challenges, we introduce a Latent Diffusion Biliary Atresia method (LDBA) to leverage imbalanced data to enhance the recognition of BA. Specifically, we propose a novel pre-trained Biliary Atresia network (PBNet) to predict labels and effectively filter out high-threshold data. Then, a new data augmentation method is proposed to focus attention on the color and anomaly points of images. Subsequently, a new label fusion method and dynamic thresholds strategy are devised to optimize the utilization of generated data by the Latent Diffusion Model (LDM). The entire process is trained in a semi-supervised manner. The experimental results indicate that our proposed LDBA outperforms competing methods with a superior performance of 78.43%. Our method can improve the accuracy and reliability of BA clinical diagnosis.

Double‐difference constrained reflection tomography in two‐dimensional elastic media

AbstractThe inverted velocity model obtained from the reflection tomography based on the angle domain common‐image gathers has a certain fuzziness. The inverted velocity model's stratigraphic interface is always not clear enough in areas with complex stratigraphic structure. In order to improve the accuracy and resolution of the inverted velocity model, a double‐difference constraint condition is added on the basis of minimizing the absolute travel‐time residual at the subsurface imaging points. This constraint makes the inverted velocity model local structure information more refined by minimizing the differential travel‐time residual at adjacent imaging points (i.e. closely spaced points within the same layer) and makes the variation of velocity model information within a certain range more accurate. The method in this paper is based on angle domain common‐image gathers, the tomography inversion equation is established by using the ray tracing method, and the conversion relationship between the traveltime residual and the residual curvature of the angle domain common‐image gathers. Then, by adding differential constraint and double‐differential constraint conditions and using the least squares QR decomposition method to solve the set of equations, the inverted velocity model can be obtained through multiple iterations, which provides a high‐precision velocity field for the migration and improves the accuracy of seismic imaging. Numerical experiments on both one typical model and a field data example demonstrate the effectiveness of the proposed double‐difference constrained elastic reflection tomography in generating high‐precision velocity models.

Mission operations management of a regional coverage highly agile SAR satellite: Few key aspects

The regional coverage satellite constellations are in demand for fulfilling earth observation needs of specific user regions of interest as opposed to global coverage thus enhancing the revisit period for the area. In this paper, we present a few key mission aspects of the regional coverage constellation designed and operated by the Indian Space Research Organization (ISRO) in low earth orbit. ISRO developed three X-Band Synthetic Aperture Radar (SAR) Earth Observation satellites and placed them in three inclined orbit planes to enhance the Indian region's coverage. This is India's first inclined orbit constellation designed for remote sensing applications in low earth orbit. The satellites were built around a standard bus carrying a Synthetic Aperture Radar payload with a deployable Radial Rib Antenna (RRA). They were designed to achieve a high agility capability. As SAR sensors are capable of all-weather, day, and night imaging; thus, most suitable in inclined orbits where the illumination conditions vary continuously for imaging. Though the inclined orbits provide more revisit opportunities to the region of interest, their characteristics pose challenges in mission design and operations. Though the mission design and planning involved a host of elements, this paper brings out the details of a few key mission analysis elements, planning aspects and operations that were critical in the course of various mission phases. The mission analysis included mission-specific requirements like launch windows selection for the constellation and reference attitude generation specific to inclined orbit geometry. Mission planning had to take care of Radial Rib Antenna (RRA) deployment requirements which were very critical for the mission, apart from taking care of individual subsystem requirements. The SAR payload data product quality and targeting accuracy are very sensitive to the satellite image point range (vis-a-vis SAR operation data window). Extensive post-launch calibration activity was carried out to meet all these mission requirements and data product specifications. The challenge was in fixing the payload beam alignment and fixing the uncertainties in onboard and ground computations. The paper also describes the methods of fixing all possible error sources affecting the high-resolution SAR payload performance.

Remarks on fixed point assertions in digital topology, 7

This paper continues a series discussing flaws in published assertions concerning fixed points in digital images.

Visual Framing of Afghan Refugees in Global News Media

This study content analyzed visual frames of Afghan refugees during the first six months of the Taliban’s second regime (from August 15, 2021, to February 15, 2022). A total of N = 1,948 images from the Associated Press were quantitatively analyzed for focal point, topic, tone, depiction, gender, and age. The findings indicate that Afghan refugees were predominantly the primary focal points of the AP images, with significantly more positive and active frames than negative and passive frames. In contrast to previous research on the framing of Afghan women in Western media before the U.S. invasion of Afghanistan, Afghan women were framed significantly as more positive and active than Afghan men in the photos.

Instance-level 6D pose estimation based on multi-task parameter sharing for robotic grasping

Six-dimensional pose estimation task predicts its 3D rotation matrix and 3D translation matrix in the world coordinate system by inputting the color image or depth image of the target object. Existing methods usually use deep neural networks to directly predict or regress object poses based on keypoint methods. The prediction results usually have deviations depending on whether the surface shape of the object is prominent or not and the size of the object. To solve this problem, we propose the six-dimensional pose estimation based on multi-task parameter sharing (PMP) framework to incorporate object category information into the pose estimation network through the form of an object classification auxiliary task. First, we extract the image features and point cloud features of the target object separately, and fuse them point by point; then, we share the confidence of each keypoint in pose estimation task and the knowledge of the classification task, get the key points with higher confidence, and predict the object pose; finally, the obtained object pose is passed through an iterative optimization network to obtain the final pose. The experimental results on the LineMOD dataset show that the proposed method can improve the accuracy of pose estimation and narrow the gap in the prediction accuracy of objects with different shapes. We also tested on a new dataset of small-scale objects, which contains object RGBD images and accurate 3D point cloud information. The proposed method is applied to the grasping experiment on the UR5 robotic arm, which satisfies the real-time pose estimation results during the grasping process.

Estimating the timber value of a forest property using geographically balanced samples and unoccupied aerial vehicle data

Abstract A common task in forestry is to determine the value of a forest property, and timber is the most valuable component of that property. Remotely sensed data collected by an unoccupied aerial vehicle (UAV) are suited for this purpose as most forest properties are of a size that permits the efficient collection of UAV data. These UAV data, when linked to a probability sample of field plots, enable the model-assisted (MA) estimation of the timber value and its associated uncertainty. Our objective was to estimate the value of timber (€/ha) in a 40-ha forest property in Finland. We used a systematic sample of field plots (n = 160) and 3D image point cloud data collected by an UAV. First, we studied the effects of spatial autocorrelation on the variance estimates associated with the timber value estimates produced using a field data-based simple expansion (EXP) estimator. The variance estimators compared were simple random sampling, Matérn, and a variant of the Grafström–Schelin estimator. Second, we compared the efficiencies of the EXP and MA estimators under different sampling intensities. The sampling intensity was varied by subsampling the systematic sample of 160 field plots. In the case of the EXP estimator, the simple random sampling variance estimator produced the largest variance estimates, whereas the Matérn estimator produced smaller variance estimates than the Grafström–Schelin estimator. The MA estimator was more efficient than the EXP estimator, which suggested that the reduction of sampling intensity from 160 to 60 plots is possible without deterioration in precision. The results suggest that the use of UAV data improves the precision of timber value estimates compared to the use of field data only. In practice, the proposed application improves the cost-efficiency of the design-based appraisal of a forest property because expensive field workload can be reduced by means of UAV data.

Frozen CLIP Transformer Is an Efficient Point Cloud Encoder

The pretrain-finetune paradigm has achieved great success in NLP and 2D image fields because of the high-quality representation ability and transferability of their pretrained models. However, pretraining such a strong model is difficult in the 3D point cloud field due to the limited amount of point cloud sequences. This paper introduces Efficient Point Cloud Learning (EPCL), an effective and efficient point cloud learner for directly training high-quality point cloud models with a frozen CLIP transformer. Our EPCL connects the 2D and 3D modalities by semantically aligning the image features and point cloud features without paired 2D-3D data. Specifically, the input point cloud is divided into a series of local patches, which are converted to token embeddings by the designed point cloud tokenizer. These token embeddings are concatenated with a task token and fed into the frozen CLIP transformer to learn point cloud representation. The intuition is that the proposed point cloud tokenizer projects the input point cloud into a unified token space that is similar to the 2D images. Comprehensive experiments on 3D detection, semantic segmentation, classification and few-shot learning demonstrate that the CLIP transformer can serve as an efficient point cloud encoder and our method achieves promising performance on both indoor and outdoor benchmarks. In particular, performance gains brought by our EPCL are 19.7 AP50 on ScanNet V2 detection, 4.4 mIoU on S3DIS segmentation and 1.2 mIoU on SemanticKITTI segmentation compared to contemporary pretrained models. Code is available at \url{https://github.com/XiaoshuiHuang/EPCL}.

Abstract 1214: Investigating therapeutic strategies to promote immune rejection of KRASG12C inhibitor-resistant subpopulations in lung cancer

Abstract Oncogenic KRAS signaling not only promotes progression and survival of cancer cells, but it can also play a major role in suppressing anti-tumor immunity through several mechanisms, including secretion of cytokines, reduction of antigen presentation and suppression of tumor-intrinsic interferon responses. Consequently, treatment with KRASG12C inhibitors alleviates oncogenic KRAS driven immune suppression, resulting in a profound remodeling of the tumor microenvironment (TME). In immunogenic mouse models, these changes can lead to the generation of some complete responses which are dependent on the engagement of the immune system. In this project we aim to investigate whether the induction of immune responses triggered by KRASG12C inhibition can effectively target inhibitor-resistant subclones. To achieve this, we mimic the development of drug resistance by combining reporter-traced isogenic KRASG12C and KRASG12D mutant lung cancer cells, where the KRASG12D mutant cells are inherently resistant to KRASG12C inhibitors. Growth of resistant cells was monitored via in vivo luciferase imaging and end point flow cytometry. KRASG12C inhibition as monotherapy, either with the active-state KRASG12C inhibitor RMC-4998 or the inactive-state inhibitor adagrasib, led to a proliferative advantage of the drug-resistant subpopulation, suggesting that any immune response to KRASG12C inhibition in the bulk tumor population was insufficient to target the resistant cell subclones. To investigate if immune mediated elimination of resistant cells could be achieved using combinations, we assessed two strategies commonly evaluated in clinical trials with KRASG12C inhibitors: anti-PD1 blockade and SHP2 inhibition. Of note, KRASG12D mutant cells do not respond to these therapies when grown subcutaneously. Treatment of RMC-4998 in combination with either anti-PD1 or the SHP2 inhibitor RMC-4550 was able to induce complete eradication of both KRASG12C and KRASG12D mutant lung cancer cells, despite the resistant cells being insensitive to these therapies when treated in the absence of sensitive cells. Furthermore, both combinations induced immune memory towards the resistant cells. These long-term effects were dependent on the immune system as they were not observed in immune deficient mice. Immune profiling of treated tumors revealed that both combinations can result in a more profound remodeling of the TME and potentiate the immune responses. Overall, our preclinical results demonstrate that the anti-tumor responses generated by the combination of KRASG12C inhibitors with SHP2 inhibition and/or immune checkpoint blockade can result in the bystander immune mediated killing of subclones of KRASG12C inhibitor resistant cells, providing a paradigm for the development of therapeutic combinations that may be more able to counter the development of inhibitor resistance. Citation Format: Mona Tomaschko, Christopher Moore, Claire E. Pillsbury, Kangbo Ng, Sophie de Carne, Sareena Rana, Andrea de Castro, Panayiotis Anastasiou, Miriam Molina-Arcas, Julian Downward. Investigating therapeutic strategies to promote immune rejection of KRASG12C inhibitor-resistant subpopulations in lung cancer [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 1214.

Split-based elevational localization of photoacoustic guidewire tip by 1D array probe using spatial impulse response.

Objective. Photoacoustic emitters on the tip of a therapeutic device have been intensively studied for echo-guided intervention purposes. In this study, a novel method for localizing the guidewire tip emitter in the elevation direction using a 1D array probe is proposed to resolve the issue of the tip potentially deviating from the ultrasound-imaged plane.Approach. Our method uses the 'interference split' that appears when the emitter is off-plane. Here, a point source from the emitter splits into two points in images. Based on the split, 'split-based elevation localization (SEL)' is introduced to estimate the absolute elevation position of the emitter. Additionally, 'Signed SEL' incorporates an asymmetric feature into the 1D probe to obtain the sign of the elevation localization. An attenuative coupler is attached to the half side of the probe to control the interference split. In SEL and Signed SEL, we propose a modeled split matching (MSM) algorithm to localize the tip position. MSM performs pattern matching of a measured split waveform with modeled split waveforms corresponding to all emitter positions in a region of interest. The modeled waveforms are precalculated using the spatial impulse response. The proposed method is numerically and experimentally validated.Main results. Numerical simulations for time-domain wave propagation clearly demonstrated the interference split phenomena. In the experimental validation with a vessel-mimicking phantom, the proposed methods successfully estimated the elevation positions,yb.SEL exhibited a root-mean-squared error (RMSE) of 2.0 mm for the range of 0 mm ≤yb≤ 30 mm, while Signed SEL estimated the absolute position with an RMSE of 2.4 mm and the sign with an accuracy of 80.8% for the range of -30 mm ≤yb≤ 30 mm.Significance.These results suggest that the proposed method could provide approximate tip positions and help sonographers track it by fanning the probe.

Sub-pixel marking and depth-based correction methods for the elimination of voxel drifting in integral imaging display.

Integral imaging is a kind of true three-dimensional (3D) display technology that uses a lens array to reconstruct vivid 3D images with full parallax and true color. In order to present a high-quality 3D image, it's vital to correct the axial position error caused by the misalignment and deformation of the lens array which makes the reconstructed lights deviate from the correct directions, resulting in severe voxel drifting and image blurring. We proposed a sub-pixel marking method to measure the axial position error of the lenses with great accuracy by addressing the sub-pixels under each lens and forming a homologous sub-pixel pair. The proposed measurement method relies on the geometric center alignment of image points, which is specifically expressed as the overlap between the test 3D voxel and the reference 3D voxel. Hence, measurement accuracy could be higher. Additionally, a depth-based sub-pixel correction method was proposed to eliminate the voxel drifting. The proposed correction method takes the voxel depth into consideration in the correction coefficient, and achieves accurate error correction for 3D images with different depths. The experimental results well confirmed that the proposed measuring and correction methods can greatly suppress the voxel drifting caused by the axial position error of the lenses, and greatly improve the 3D image quality.

Virtual point tracking method for online detection of relative wheel-rail displacement of railway vehicles

Relative wheel-rail displacement (RWRD) is an important physical quantity that responds to the hunting stability and running safety of railway vehicles but detecting this physical quantity is challenging during train operation due to the complex service environment and the nonlinearity of vehicle-rail system. To address this problem, this paper proposes a virtual point tracking (VPT) method to detect RWRD. First, a series of regions of interest (ROIs) in the preprocessed wheel-rail contact images are extracted using a simplified YOLO (SYOLO) model. Then, some key positions of the wheel and rail on the extracted ROIs are identified using an improved UNet (IUNet) model and characterized by virtual points. Finally, the global coordinates of the virtual points in the whole wheel-rail contact image are calculated to obtain the final RWRD. To demonstrate the feasibility of the VPT method, it is used to detect the RWRD of a vehicle operating on a full-scale roller rig. For comparison, the original UNet model is directly used to detect the RWRD. Two quantities, Mean Absolute Error (MAE) and Root Mean Square Error (RMSE), are applied for quantitative analysis and the results show that the VPT method can identify the RWRD more reliably and accurately.

A new strategy for prestack time migration velocity analysis to assess and mitigate structural uncertainty

Velocity uncertainty represents a significant contributor to structural uncertainty in seismic imaging. Therefore, improving velocities and enhancing the resolution of velocity spectra are crucial for reducing structural uncertainty. In the present study, a novel approach is introduced to increase velocity resolution and mitigate structural uncertainty. The methodology's impact on spatial and temporal uncertainty of image points is compared with the improved velocity continuation method. The proposed strategy involves modifying the equations of the weighted semblance method and integrating them into the equations of the velocity continuation method. Specifically, the vertical analysis window of the weighted semblance, typically applied on coherency panels is changed to a horizontal window that is applied on common-image gathers. Correspondingly, the weighting factor is adjusted using a time-adaptive function. The velocity uncertainty equation in the velocity continuation method is then replaced by the new weighted semblance eq. A similar substitution can be performed for structural uncertainty equations. These new equations are then used on common-image gathers obtained by the velocity continuation method to increase the accuracy of the velocity model and then investigate changes in structural uncertainty. In the proposed strategy, prestack time migration is initially conducted on seismic common-offset sections using various constant velocities within a predefined range. Subsequently, the weighted semblance technique is used for automatic velocity model updating in all common-image gathers. The new velocity model is then used for final prestack time migration. In the equations pertaining to velocity uncertainty, a width equivalent to one standard deviation on both sides of the picked velocity value is considered as the uncertainty range. By narrowing this width through the proposed method, it is expected to mitigate subsequent structural uncertainty, specifically for 1D media. Results from applying the presented method on synthetic and marine seismic data demonstrate its effectiveness in reducing structural uncertainty for 1D time migrated images compared to the velocity continuation method. It is also shown that horizontal structural uncertainty is significantly more sensitive to velocity errors rather than vertical uncertainty. An added benefit is that the increased resolution in velocity spectra also provides appropriate panels for automatic velocity picking.

Scattering-Point-Guided Oriented RepPoints for Ship Detection

Ship detection finds extensive applications in fisheries management, maritime rescue, and surveillance. However, detecting nearshore targets in SAR images is challenging due to land scattering interference and non-axisymmetric ship shapes. Existing SAR ship detection models struggle to adapt to oriented ship detection in complex nearshore environments. To address this, we propose an oriented-reppoints target detection scheme guided by scattering points in SAR images. Our method deeply integrates SAR image target scattering characteristics and designs an adaptive sample selection scheme guided by target scattering points. This incorporates scattering position features into the sample quality measurement scheme, providing the network with a higher-quality set of proposed reppoints. We also introduce a novel supervised guidance paradigm that uses target scattering points to guide the initialization of reppoints, mitigating the influence of land scattering interference on the initial reppoints quality. This achieves adaptive feature learning, enhancing the quality of the initial reppoints set and the performance of object detection. Our method has been extensively tested on the SSDD and HRSID datasets, where we achieved mAP scores of 89.8% and 80.8%, respectively. These scores represent significant improvements over the baseline methods, demonstrating the effectiveness and robustness of our approach. Additionally, our method exhibits strong anti-interference capabilities in nearshore detection and has achieved state-of-the-art performance.

Enhancing Single-Plane Fluoroscopy: A Self-Calibrating Bundle Adjustment for Distortion Modeling.

Single-plane fluoroscopy systems with image intensifiers remain commonly employed in a clinical setting. The imagery they capture is vulnerable to several types of geometric distortions introduced by the system's components and their assembly as well as interactions with the local and global magnetic fields. In this study, the application of a self-calibrating bundle adjustment is investigated as a method to correct geometric distortions in single-plane fluoroscopic imaging systems. The resulting calibrated imagery is then applied in the quantitative analysis of diaphragmatic motion and potential diagnostic applications to hemidiaphragm paralysis. The calibrated imagery is further explored and discussed in its potential impact on areas of surgical navigation. This work was accomplished through the application of a controlled experiment with three separate Philips Easy Diagnost R/F Systems. A highly redundant (~2500 to 3500 degrees-of-freedom) and geometrically strong network of 18 to 22 images of a low-cost target field was collected. The target field comprised 121 pre-surveyed tantalum beads embedded on a 25.4 mm × 25.4 mm acrylic base plate. The modeling process resulted in the estimation of five to eight distortion coefficients, depending on the system. The addition of these terms resulted in 83-85% improvement in terms of image point precision (model fit) and 85-95% improvement in 3D object reconstruction accuracy after calibration. This study demonstrates significant potential in enhancing the accuracy and reliability of fluoroscopic imaging, thereby improving the overall quality and effectiveness of medical diagnostics and treatments.

Relative vectoring using dual object detection for autonomous aerial refueling

AbstractOnce realized, autonomous aerial refueling will revolutionize unmanned aviation by removing current range and endurance limitations. Previous attempts at establishing vision-based solutions have come close but rely heavily on near perfect extrinsic camera calibrations that often change midflight. In this paper, we propose dual object detection, a technique that overcomes such requirement by transforming aerial refueling imagery directly into receiver aircraft reference frame probe-to-drogue vectors regardless of camera position and orientation. These vectors are precisely what autonomous agents need to successfully maneuver the tanker and receiver aircraft in synchronous flight during refueling operations. Our method follows a common 4-stage process of capturing an image, finding 2D points in the image, matching those points to 3D object features, and analytically solving for the object pose. However, we extend this pipeline by simultaneously performing these operations across two objects instead of one using machine learning and add a fifth stage that transforms the two pose estimates into a relative vector. Furthermore, we propose a novel supervised learning method using bounding box corrections such that our trained artificial neural networks can accurately predict 2D image points corresponding to known 3D object points. Simulation results show that this method is reliable, accurate (within 3 cm at contact), and fast (45.5 fps).

FFLDGA-Net: Image retrieval method based on Feature Fusion Learnable Descriptor Graph Attention Network

Image retrieval aims to retrieve and return the image in the database that is most similar to the query image. However, the performance of image retrieval models is often hindered by the limited dimensionality of images, which lacks depth information about objects. To address this issue, we propose a novel image retrieval model called FFLDGA-Net (Feature Fusion-based Learnable Descriptor Graph Attention Network). This model aims to overcome the absence of depth information in images by fusing feature information from both image data and point cloud data. First, we introduce the LDGA-Net, which effectively improved the model’s ability to mine hard samples and negative samples. Then, we combine the multi-scale route convolution module with a one-dimensional path aggregation network to fuse point clouds and image features at multiple scales, and establish a high-dimensional relationship between features and low-dimensional features. To mitigate training noise, we incorporate a soft label strategy tailored to the dataset’s characteristics. Our experimental results on two benchmark datasets demonstrate the significant improvements achieved by FFLDGA-Net in image retrieval performance.

A Coarse-to-Fine Transformer-Based Network for 3D Reconstruction from Non-Overlapping Multi-View Images

Reconstructing 3D structures from non-overlapping multi-view images is a crucial task in the field of 3D computer vision, since it is difficult to establish feature correspondences and infer depth from overlapping parts of views. Previous methods, whether generating the surface mesh or volume of an object, face challenges in simultaneously ensuring the accuracy of detailed topology and the integrity of the overall structure. In this paper, we introduce a novel coarse-to-fine Transformer-based reconstruction network to generate precise point clouds from multiple input images at sparse and non-overlapping viewpoints. Specifically, we firstly employ a general point cloud generation architecture enhanced by the concept of adaptive centroid constraint for the coarse point cloud corresponding to the object. Subsequently, a Transformer-based refinement module applies deformation to each point. We design an attention-based encoder to encode both image projection features and point cloud geometric features, along with a decoder to calculate deformation residuals. Experiments on ShapeNet demonstrate that our proposed method outperforms other competing methods.

Perspective deformation correction for circular pointer meter based on prior indication structure feature

The meter verification and industrial inspection robots often use automatic reading of pointer meters. Due to tilt and rotation of circular pointer meter, the perspective deformation may be induced to the captured image, so it will extremely affect the accuracy of automatic reading algorithm based on machine vision. Some existing deformation correction methods suffer from poor universality or low accuracy. The paper bases on prior indication structure feature of circular pointer meter and presents a new deformation correction method. First, the MobileNetV3-UNet network is used to detect the outer-endpoints of long scale lines in deformed meter as the key points. Second, with the help of prior information, the positions of mapping points in standard image are constructed. Third, according to the one-to-one mapping relationship of key points from deformed and standard images, a perspective transform matrix is calculated. Finally, the deformation and rotation correction are performed through inverse perspective transform. The experimental results show that the proposed method has better precision and robustness and works well under different conditions.

Generating a full spherical view by modeling the relation between two fisheye images

Full spherical views provide advantages in many applications that use visual information. Dual back-to-back fisheye cameras are receiving much attention to obtain this type of view. However, obtaining a high-quality full spherical view is very challenging. In this paper, we propose a correction step that models the relation between the pixels of the pair of fisheye images in polar coordinates. This correction is implemented during the mapping from the unit sphere to the fisheye image using the equidistant fisheye projection. The objective is that the projections of the same point in the pair of images have the same position on the unit sphere after the correction. In this way, they will also have the same position on the equirectangular coordinate system. Consequently, the discontinuity between the spherical views for blending is minimized. Throughout the manuscript, we show that the angular polar coordinates of the same scene point in the fisheye images are related by a sine function and the radial distance coordinates by a linear function. Also, we propose employing a polynomial as a geometric transformation between the pair of spherical views during the image alignment since the relationship between the matching points of pairs of spherical views is not linear, especially in the top/bottom regions. Quantitative evaluations demonstrate that using the correction step improves the quality of the full spherical view, i.e. IQ MS-SSIM, up to 7%. Similarly, using a polynomial improves the IQ MS-SSIM up to 6.29% with respect to using an affine matrix.