Image Detail Information Research Articles

MS-Pansharpening Algorithm Based on Dual Constraint Guided Filtering

The difference and complementarity of spatial and spectral information between multispectral (MS) image and panchromatic (PAN) image have laid the foundation for the fusion of the two types of images. In recent years, MS and PAN image fusion (also known as MS-Pansharpening) has gained attention as an important research area in remote sensing (RS) image processing. This paper proposes an MS-Pansharpening algorithm based on dual constraint Guided Filtering in the nonsubsampled shearlet transform (NSST) domain. The innovation is threefold. First, the dual constraint guided image filtering (DCGIF) model, based on spatial region average gradient correlation and vector correlation formed by neighborhood elements is proposed. Further, the PAN image detail information extraction scheme, based on the model, is provided, which extracts more complete and accurate detail information, thus avoiding, to some extent, the spectral distortion caused by the injection of non-adaptive information. Second, the weighted information injection model, based on the preservation of the correlation between the band spectra, is proposed. The model determines the information injection weight of each band pixel based on the spectral proportion between bands of the original MS image, which ensures the spectral correlation between bands of the fused MS image. Finally, a new MS-Pansharpening algorithm in NSST domain is proposed. The MS and PAN high frequency sub-bands of NSST are used to extract more effective spatial details. Then the proposed DCGIF model is used to extract the effective spatial detail injection information through the weighted joint method based on the regional energy matrix. Finally, the weighted information injection model is used to inject it into each band of MS to complete information fusion. Experimental results show that the proposed approach has better fusion effect than some conventional MS-Pansharpening algorithms, which can effectively improve the spatial resolution of the fused MS image and maintain the spectral characteristics of MS.

Application of three-dimensional computed tomography bronchography and angiography in thoracoscopic anatomical segmentectomy of the right upper lobe: A cohort study.

ObjectiveThree-dimensional computed tomography bronchography and angiography (3D-CTBA) can provide detailed imaging information for pulmonary segmentectomy. This study aimed to investigate the safety and effectiveness of 3D-CTBA guidance of anatomical segmentectomy of the right upper lobe (RUL).MethodsThis was a retrospective analysis of anatomical segmentectomy of the RUL at the Thoracic Surgery Department of the Fourth Hospital of Hebei Medical University from December 9, 2013, to June 2, 2021. Preoperatively, all patients underwent contrast-enhanced CT of the chest (to determine the size of the pulmonary nodule) and a lung function test. 3D-CTBA has been performed since 2018; patients with vs. without 3D-CTBA were compared. Segmentectomy was performed according to nodule location.ResultsOf 139 patients (46 males and 93 females, aged 21–81 years), 93 (66.9%) completed single segmentectomy, 3 (2.2%) completed single subsegmentectomy, 29 had combined subsegmentectomy, 7 had segmentectomy combined with subsegmentectomy, and 6 had combined resection of two segments. Eighty-five (61.2%) patients underwent 3D-CTBA. 3D-CTBA cases had decreased intraoperative blood loss (67.4 ± 17.6 vs. 73.1 ± 11.0, P = 0.021) and shorter operation time (143.0 ± 10.8 vs. 133.4 ± 20.9, P = 0.001). 3D-CTBA (Beta = −7.594, 95% CI: −12.877 to −2.311, P = 0.005) and surgical procedure (Beta = 9.352, 95% CI: 3.551–15.153, P = 0.002) were independently associated with intraoperative blood loss. 3D-CTBA (Beta = −13.027, 95% CI: −18.632 to 17.422, P < 0.001) and surgical procedure (Beta = 7.072, 95% CI: 0.864–13.280, P = 0.026) were also independent factors affecting the operation time.ConclusionPreoperative use of 3D-CTBA to evaluate the pulmonary vessels and bronchial branch patterns of the RUL decreased blood loss and procedure time and so would be expected to improve the safety and effectiveness of thoracoscopic segmentectomy.

Scale-Space Feature Recalibration Network for Single Image Deraining.

Computer vision technology is increasingly being used in areas such as intelligent security and autonomous driving. Users need accurate and reliable visual information, but the images obtained under severe weather conditions are often disturbed by rainy weather, causing image scenes to look blurry. Many current single image deraining algorithms achieve good performance but have limitations in retaining detailed image information. In this paper, we design a Scale-space Feature Recalibration Network (SFR-Net) for single image deraining. The proposed network improves the image feature extraction and characterization capability of a Multi-scale Extraction Recalibration Block (MERB) using dilated convolution with different convolution kernel sizes, which results in rich multi-scale rain streaks features. In addition, we develop a Subspace Coordinated Attention Mechanism (SCAM) and embed it into MERB, which combines coordinated attention recalibration and a subspace attention mechanism to recalibrate the rain streaks feature information learned from the feature extraction phase and eliminate redundant feature information to enhance the transfer of important feature information. Meanwhile, the overall SFR-Net structure uses dense connection and cross-layer feature fusion to repeatedly utilize the feature maps, thus enhancing the understanding of the network and avoiding gradient disappearance. Through extensive experiments on synthetic and real datasets, the proposed method outperforms the recent state-of-the-art deraining algorithms in terms of both the rain removal effect and the preservation of image detail information.

Improved recurrent generative model for reconstructing large-size porous media from two-dimensional images.

Modeling the three-dimensional (3D) structure from a given 2D image is of great importance for analyzing and studying the physical properties of porous media. As an intractable inverse problem, many methods have been developed to address this fundamental problems over the past decades. Among many methods, the deep learning-(DL) based methods show great advantages in terms of accuracy, diversity, and efficiency. Usually, the 3D reconstruction from the 2D slice with a larger field-of-view is more conducive to simulate and analyze the physical properties of porous media accurately. However, due to the limitation of reconstruction ability, the reconstruction size of most widely used generative adversarial network-based model is constrained to 64^{3} or 128^{3}. Recently, a 3D porous media recurrent neural network based method (namely, 3D-PMRNN) (namely 3D-PMRNN) has been proposed to improve the reconstruction ability, and thus the reconstruction size is expanded to 256^{3}. Nevertheless, in order to train these models, the existed DL-based methods need to down-sample the original computed tomography (CT) image first so that the convolutional kernel can capture the morphological features of training images. Thus, the detailed information of the original CT image will be lost. Besides, the 3D reconstruction from a optical thin sectionis not available because of the large size of the cutting slice. In this paper, we proposed an improved recurrent generative model to further enhance the reconstruction ability (512^{3}). Benefiting from the RNN-based architecture, the proposed model requires only one 3D training sample at least and generates the 3D structures layer by layer. There are three more improvements: First, a hybrid receptive field for the kernel of convolutional neural network is adopted. Second, an attention-based module is merged into the proposed model. Finally, a useful sectionloss is proposed to enhance the continuity along the Z direction. Three experiments are carried out to verify the effectiveness of the proposed model. Experimental results indicate the good reconstruction ability of proposed model in terms of accuracy, diversity, and generalization. And the effectiveness of sectionloss is also proved from the perspective of visual inspection and statistical comparison.

An Image Deblurring Method Using Improved U-Net Model

Deblurring methods in dynamic scenes are a challenging problem. Recently, significant progress has been made for image deblurring methods based on deep learning. However, these methods usually stack ordinary convolutional layers or increase convolution kernel size, resulting in limited receptive fields, an unsatisfying deblurring effect, and a heavy computational burden. Therefore, we propose an improved U-Net (U-shaped Convolutional Neural Network) model to restore the blurred images. We first design the model structure, which mainly includes depth-wise separable convolution, residual depth-wise separable convolution, wavelet transform, inverse wavelet transform, and a DMRFC (dense multireceptive field channel) module. Next, a depth-wise separable convolution is designed, which reduces model calculations and the number of parameters when compared with the standard convolution. A residual depth-wise separable convolution is designed, which allows for propagation of detailed information from different layers when compared with standard convolution and a standard residual block. The wavelet transform realizes downsampling by separating the contextual and texture information of the image. It also reduces model training difficulty. The inverse wavelet transform realizes upsampling, which reduces the loss of image information. Finally, by combining an extensional receptive field and channel attention mechanism, a DMRFC module is proposed to extract detailed image information, which further improves the reconstructed image quality via inverse wavelet transform. Experiments on the public dataset GOPRO show that the image deblurring method in this paper has higher-quality visual effects, while the PSNR (peak signal-to-noise ratio) rises to 30.83 and SSIM (structural similarity) rises to 0.948. Fewer model parameters and a shorter recovery time are needed, which provides a more lightweight image deblurring method.

Infrared and visible image fusion using multi-scale pyramid network

Deep networks have been widely applied in infrared and visible image fusion. However, the current deep networks cannot well extract and fuse multi-scale information and high-frequency texture features of the source images. In this paper, a deep multi-scale pyramid network, termed MSPFNet, is proposed for infrared and visible image fusion by combining image Laplacian pyramid and deep network. Infrared and visible images are first decomposed into their Laplacian pyramids. For each source image, its Laplacian pyramid consists of a low-frequency component and a series of multi-scale high-frequency components containing texture details. Then, the Laplacian pyramid components of two source images in the same level are fused using convolutional neural networks (CNN). Finally, the final fused image is reconstructed on the fused Laplacian pyramid components using inverse Laplacian pyramid transform. The experimental results on publicly available datasets show that MSPFNet can efficiently extract and fuse the multi-scale detail information of source images, and the fused images of MSPFNet preserve more texture details of infrared and visible images than that of the previous state-of-the-art methods.

Performance Improvement of Face Recognition Method and its Application in COVID-19

In this paper an efficient face recognition technique is presented by integrating Discrete Wavelet Transform and Compressive Sensing based classifier. At first discrete wavelet transform has been applied on each face images. Then an image fusion technique has been applied on the decomposed image to provide better detail information of face images. Principal component Analysis is applied on fused face images to extract the feature vector. Finally, the feature vector of test images is extracted and classified by Compressive Sensing based Classifier. This proposed technique is also tested on two publicly available databases on AR and ORL. This technique is also tested on masked face images and experimental result shows improved performance compared to conventional PCA.

MCPANet: Multiscale Cross-Position Attention Network for Retinal Vessel Image Segmentation

Accurate medical imaging segmentation of the retinal fundus vasculature is essential to assist physicians in diagnosis and treatment. In recent years, convolutional neural networks (CNN) are widely used to classify retinal blood vessel pixels for retinal blood vessel segmentation tasks. However, the convolutional block receptive field is limited, simple multiple superpositions tend to cause information loss, and there are limitations in feature extraction as well as vessel segmentation. To address these problems, this paper proposes a new retinal vessel segmentation network based on U-Net, which is called multi-scale cross-position attention network (MCPANet). MCPANet uses multiple scales of input to compensate for image detail information and applies to skip connections between encoding blocks and decoding blocks to ensure information transfer while effectively reducing noise. We propose a cross-position attention module to link the positional relationships between pixels and obtain global contextual information, which enables the model to segment not only the fine capillaries but also clear vessel edges. At the same time, multiple scale pooling operations are used to expand the receptive field and enhance feature extraction. It further reduces pixel classification errors and eases the segmentation difficulty caused by the asymmetry of fundus blood vessel distribution. We trained and validated our proposed model on three publicly available datasets, DRIVE, CHASE, and STARE, which obtained segmentation accuracy of 97.05%, 97.58%, and 97.68%, and Dice of 83.15%, 81.48%, and 85.05%, respectively. The results demonstrate that the proposed method in this paper achieves better results in terms of performance and segmentation results when compared with existing methods.

MDIVis: Visual analytics of multiple destination images on tourism user generated content

Abundant tourism user-generated content (UGC) contains a wealth of cognitive and emotional information, providing valuable data for building destination images that depict tourists’ experiences and appraisal of the destinations during the tours. In particular, multiple destination images can assist tourism managers in exploring the commonalities and differences to investigate the elements of interest of tourists and improve the competitiveness of the destinations. However, existing methods usually focus on the image of a single destination, and they are not adequate to analyze and visualize UGC to extract valuable information and knowledge. Therefore, we discuss requirements with tourism experts and present MDIVis, a multi-level interactive visual analytics system that allows analysts to comprehend and analyze the cognitive themes and emotional experiences of multiple destination images for comparison. Specifically, we design a novel sentiment matrix view to summarize multiple destination images and improve two classic views to analyze the time-series pattern and compare the detailed information of images. Finally, we demonstrate the utility of MDIVis through three case studies with domain experts on real-world data, and the usability and effectiveness are confirmed through expert interviews.

Convolutional block attention module U-Net: a method to improve attention mechanism and U-Net for remote sensing images

Building extraction from high-resolution remote sensing images is the foundation of many fields, but the existing algorithms cannot extract building details from remote sensing images well because of complex background conditions and the occlusion of interfering objects. To solve the problems of low accuracy and incomplete boundary of traditional building extraction methods, ProCBAM, a parallel attention mechanism based on U-Net network, is adopted and added to the feature transmission step of U-Net. Through the representation of spatial dimension image feature map and channel dimension optimized feature map, we can learn more detailed image information and reduce the error of image recognition. Experiments are carried out on the Massachusetts building dataset, Wuhan University dataset, and IND.v2 dataset, and the experimental results show the effectiveness of this method in building extraction from remote sensing images.

Using global information to refine local patterns for texture representation and classification

Local binary pattern (LBP) and its variants have been successfully applied in texture feature extraction. However, it is hard for most LBP-based methods to effectively describe and distinguish the local neighborhoods with similar structures (that is, the calculated feature patterns are identical) but different contrasts or grayscales. To alleviate such problems, we propose a novel global refined local binary pattern (GRLBP) by analyzing the nature of pixel intensity distribution in local neighborhoods. GRLBP consists of two descriptors called magnitude refined local sign binary pattern (MRLBP_S) and center refined local magnitude binary pattern (CRLBP_M). MRLBP_S distinguishes local neighborhoods with contrast differences by using global magnitude anchors to refine local sign patterns. And CRLBP_M identifies local neighborhoods with grayscale differences by employing global central grayscale anchors to refine local magnitude patterns. Finally, frequency histograms of MRLBP_S and CRLBP_M from each image are cascaded to generate the GRLBP. Extensive experimental results on seven benchmark texture databases: Outex, CUReT, KTH-TIPS, UMD, UIUC, KTH-T2b, and DTD demonstrate that the proposed GRLBP can represent the detailed information of texture images. Furthermore, compared with state-of-the-art LBP variants, GRLBP has competitive advantages in classification accuracy, feature dimension, and computational complexity, respectively.

Low Light Image Enhancement Algorithm Based on Detail Prediction and Attention Mechanism.

Most LLIE algorithms focus solely on enhancing the brightness of the image and ignore the extraction of image details, leading to losing much of the information that reflects the semantics of the image, losing the edges, textures, and shape features, resulting in image distortion. In this paper, the DELLIE algorithm is proposed, an algorithmic framework with deep learning as the central premise that focuses on the extraction and fusion of image detail features. Unlike existing methods, basic enhancement preprocessing is performed first, and then the detail enhancement components are obtained by using the proposed detail component prediction model. Then, the V-channel is decomposed into a reflectance map and an illumination map by proposed decomposition network, where the enhancement component is used to enhance the reflectance map. Then, the S and H channels are nonlinearly constrained using an improved adaptive loss function, while the attention mechanism is introduced into the algorithm proposed in this paper. Finally, the three channels are fused to obtain the final enhancement effect. The experimental results show that, compared with the current mainstream LLIE algorithm, the DELLIE algorithm proposed in this paper can extract and recover the image detail information well while improving the luminance, and the PSNR, SSIM, and NIQE are optimized by 1.85%, 4.00%, and 2.43% on average on recognized datasets.

Application of Addition and Multiplication Noise Model Parameter Estimation in INSAR Image Processing

INSAR images are inevitably contaminated by noise during the process of generation, transmission, compression, and reception. Noise not only affects the quality of the INSAR image, but also affects subsequent operations such as the design of corresponding filters, INSAR image segmentation, compression, restoration, and feature recognition. The INSAR image noise model is mainly divided into additive noise and multiplicative noise. Compared with additive noise, multiplicative noise is more complicated due to INSAR image correlation and non-Gaussian. Based on least squares algorithm system of additive and multiplicative mixed noise model, this paper proposes a method of using PCA to remove multiplicative gamma distribution noise. The pure noise coefficient is obtained by subtracting the original coefficient from the diagonal wavelet coefficient of the noisy image, and the mode of the local variance is calculated as the estimation value of the noise standard deviation. Experiments show that the proposed method can obtain more accurate estimation of noise; in particular in the case of less noise and more detailed image information, its effect is more obvious.

MRCON-Net: Multiscale reweighted convolutional coding neural network for low-dose CT imaging

Background and objectiveLow-dose computed tomography (LDCT) has become increasingly important for alleviating X-ray radiation damage. However, reducing the administered radiation dose may lead to degraded CT images with amplified mottle noise and nonstationary streak artifacts. Previous studies have confirmed that deep learning (DL) is promising for improving LDCT imaging. However, most DL-based frameworks are built intuitively, lack interpretability, and suffer from image detail information loss, which has become a general challenging issue. MethodsA multiscale reweighted convolutional coding neural network (MRCON-Net) is developed to address the above problems. MRCON-Net is compact and more explainable than other networks. First, inspired by the learning-based reweighted iterative soft thresholding algorithm (ISTA), we extend traditional convolutional sparse coding (CSC) to its reweighted convolutional learning form. Second, we use dilated convolution to extract multiscale image features, allowing our single model to capture the correlations between features of different scales. Finally, to automatically adjust the elements in the feature code to correct the obtained solution, a channel attention (CA) mechanism is utilized to learn appropriate weights. ResultsThe visual results obtained based on the American Association of Physicians in Medicine (AAPM) Challenge and United Image Healthcare (UIH) clinical datasets confirm that the proposed model significantly reduces serious artifact noise while retaining the desired structures. Quantitative results show that the average structural similarity index measurement (SSIM) and peak signal-to-noise ratio (PSNR) achieved on the AAPM Challenge dataset are 0.9491 and 40.66, respectively, and the SSIM and PSNR achieved on the UIH clinical dataset are 0.915 and 42.44, respectively; these are promising quantitative results. ConclusionCompared with recent state-of-the-art methods, the proposed model achieves subtle structure-enhanced LDCT imaging. In addition, through ablation studies, the components of the proposed model are validated to achieve performance improvements.

An advanced algorithm for fusing Gaofen multispectral satellite data with drone imagery

ABSTRACT Common image fusion methods used to merge remotely sensed data of varying spatial and spectral information as provided by different air-, and space-borne sensors often do not provide the optimal reachable information. This paper proposes a novel fusion method for pansharpening using a multi-resolution analysis (MRA) based on Chinese Gaofen (GF-1/2) satellite data and corresponding drone data as an example. The novel fusion method is named droesharpening fusion method, which addresses common poor spatial and spectral feature issues of fusion results. The droesharpening fusion method designed several separate computing steps based on wavelets and a pulse-coupled neural network (PCNN) to calculate low-resolution drone (LDRO) images and an injection weight (g k ) for significantly improved results. The droesharpening fusion method includes a new algorithm for reducing the MDRO image dimensions (RDA). The proposed new algorithm calculated image feature (IFA), and the proposed algorithm of calculating the gk (IWA). ARD is used to reduce the dimensionalities of a drone (MDRO) image to a new single-band drone (SDRO) image. Then, we applied an improved ‘à trous’ Wavelet Transform (ATWT) to SDRO images to extract LDRO images and an improved PCNN for the multispectral (MS) images. Additionally, edge processing is deployed to the PCNN segmented images. The IFA is then used to calculate the spatial characteristics of the SDRO images based on the ATWT decomposed image and the spectral features of the MS images based on the PCNN segmented images. In the next step, the IMA calculates the gk based on the spatial and spectral characteristics. The detailed spatial information of the MDRO images is then injected into the MS images by using the gk to obtain the fused drone/multispectral satellite imagery. Additional experimental results using several drone/multispectral satellite images proved that the proposed method achieves high spectral fidelity and simultaneously improves the spatial resolution of fused images.

MFFE: Multi-scale Feature Fusion Enhanced Net for image dehazing

It is very challenging to perform dehazing operations on single haze images taken due to information degradation, for which we propose a novel single-image dehazing network based on U-Net. Most existing image dehazing algorithms only focus on whether they can remove the haze, but ignore the quality of the final dehazed image, which can lead to problems such as loss of image detail information and texture blurring, affecting the final recovery. We use the enhancement and error feedback mechanisms inside the super-resolution algorithm to gradually recover the haze-free images by introducing the Strengthen-Operate-subtract strategy in the decoder and demonstrate their effectiveness for the dehazing problem. Meanwhile, to solve the problem of missing information and utilization in the traditional U-Net structure, we design a multiscale feature fusion module that can effectively compensate for the missing contextual information and make full use of the disjoint features. In addition, an improved local binary pattern and SE attention mechanism are used to help the network obtain clearer details and texture images, and the accuracy of the enhanced dehazing network is improved using residual learning. We analyze the effectiveness of the proposed algorithm and show through extensive evaluation that our proposed model can be effectively used for single image dehazing and also has good performance compared to state-of-the-art methods on the test dataset.

Momentum feature comparison network based on generative adversarial network for single image super-resolution

Most super-resolution methods are trained on datasets where high-resolution images and corresponding low-resolution images are obtained by the fixed degradation method. However, these external-based methods would fail to recover the detailed information of the test images if it could not be found in datasets. In this paper, we propose the momentum feature comparison network to generate the super-resolution image with rich texture information. Without the participation of high-resolution images in the training process, our method belongs to the completely unsupervised super-resolution method. A siamese structure with momentum update in generator is proposed to expand the content information of the low-resolution image and maintains the continuous consistency of intermediate feature maps. Furthermore, the results of two branches are fused through the feature fusion module to retain the global distribution of features and enhance local high-frequency details. The experimental results show that our method achieves great results compared with the state-of-the-art methods.

MSAt-GAN: a generative adversarial network based on multi-scale and deep attention mechanism for infrared and visible light image fusion

For the past few years, image fusion technology has made great progress, especially in infrared and visible light image infusion. However, the fusion methods, based on traditional or deep learning technology, have some disadvantages such as unobvious structure or texture detail loss. In this regard, a novel generative adversarial network named MSAt-GAN is proposed in this paper. It is based on multi-scale feature transfer and deep attention mechanism feature fusion, and used for infrared and visible image fusion. First, this paper employs three different receptive fields to extract the multi-scale and multi-level deep features of multi-modality images in three channels rather than artificially setting a single receptive field. In this way, the important features of the source image can be better obtained from different receptive fields and angles, and the extracted feature representation is also more flexible and diverse. Second, a multi-scale deep attention fusion mechanism is designed in this essay. It describes the important representation of multi-level receptive field extraction features through both spatial and channel attention and merges them according to the level of attention. Doing so can lay more emphasis on the attention feature map and extract significant features of multi-modality images, which eliminates noise to some extent. Third, the concatenate operation of the multi-level deep features in the encoder and the deep features in the decoder are cascaded to enhance the feature transmission while making better use of the previous features. Finally, this paper adopts a dual-discriminator generative adversarial network on the network structure, which can force the generated image to retain the intensity of the infrared image and the texture detail information of the visible image at the same time. Substantial qualitative and quantitative experimental analysis of infrared and visible image pairs on three public datasets show that compared with state-of-the-art fusion methods, the proposed MSAt-GAN network has comparable outstanding fusion performance in subjective perception and objective quantitative measurement.

An Effective Approach for Sonar Image Recognition with Improved Efficientdet and Ensemble Learning

Abstract Precisely recognition of underwater targets plays a critical role in the field of underwater unmanned exploration. According to the characteristics of sonar image, in order to improve the accuracy of underwater automatic target recognition, this paper proposes an underwater target sonar image recognition method combined with improved Efficientdet and ensemble learning. According to the regional dominance principle and threshold segmentation method, the sonar image is preprocessed, and preliminary effective features extracted from preprocessed data by Efficientnet backbone, so as to balance the efficiency and power consumption of the algorithm. The feature fusion adopts the improved weighted bi-directional feature pyramid ( BiFPN ) structure to aggregate the global information and strengthen the feature representation of the shallow feature map, which optimizes the problem that the small feature extraction is not ideal due to the low resolution of the sonar image and less detail information. By building a two-level classifier to complete the learning task in an ensemble way, a better classification effect is obtained than a single classifier. This method is applied to the underwater intelligent perception group of the 10th National Marine Aircraft Design and Production Competition in 2021, and won the first prize in the country. The experimental results show that the designed ensemble network model is more accurate than the conventional convolution neural network in identifying underwater targets on the measured sonar image data set while ensuring a certain prediction speed.

Deep learning-based stereo matching using the feature spatial pyramid pooling

For deep learning-based stereo matching, despite deep convolution networks possessing rich, high-level semantic features that are usually abstract and low-resolution, it is difficult to accurately calculate per-pixel correspondence without image detail information in challenging regions, such as thin objects, edge contours, reflective surfaces, etc. Motivated by the information propagation capability of the feature pyramid network (FPN), we focus on using FPN to improve the detailed comprehension of spatial pyramid pooling (SPP). We propose a feature SPP (FSPP), which consists of FPN and SPP. FSPP integrates high-frequency details reserved by FPN into multiscale features extracted by SPP to form a feature volume considering both detailed and contextual information. Using FSPP, features become distinct and alleviate the ambiguity of matching progress in challenging regions. Furthermore, we concatenate FSPP to Pyramid Stereo Matching Network with three-dimensional convolution and hourglass architecture. We also evaluate its effectiveness on the KITTI 2012/2015 and the Middlebury 2014/2021 datasets. The results show that the proposed algorithm performs better than current state-of-the-art algorithms in challenging regions while taking on better generalization.