Infrared and visual image fusion aims to integrate the salient and complementary features of the infrared image and visual image into one informative image. To achieve this purpose, we have proposed an infrared and visual image fusion method via iterative quadtree decomposition and Bézier interpolation. To be specific, each source image is first decomposed to image patches of multiple sizes in a quadtree structure according to a fixed threshold, then each image patch in the quadtree structure is smoothed by interpolating its four-by-four uniformly distributed pixels with the Bézier interpolation method. With the iteratively smoothed images, multiple scales of bright and dark feature maps of each source image can be gradually extracted from the difference image of every two continuously smoothed images. At last, fusion of the infrared image and visual image can be realized by fusing their multiple scales of bright, dark features and their base images (i.e., final-scale smoothed images). Extensive experiments verify that the proposed method outperforms five state-of-the-art image fusion methods in both qualitative and quantitative evaluations.
Small ship detection in remote sensing images is an increasingly important area in computer vision. The main greatest challenges that face it are a small size and different aspect ratios of ships. Recently, convolution neural network (CNN) achieve outstanding performance in object detection. CNN consists of several consecutive layers categorized to shallow and deep according to their positions. The shallow layers contribute in the detection of tiny targets significantly. However, the semantic features of these layers are weak to classify them correctly. In this paper, an aggregation context network is proposed to enhance the semantic features of the shallow layers for the state-of-the-art RetinaNet detector. This network is located before the feature pyramid network of the RetinaNet. It consists of aggregation and context modules. The aggregation module aggregates the different features to improve the semantic features at the shallow layers. The context module is to increase the receptive fields at each layer of the feature pyramid network. The experiments are carried on a proposed ship dataset. The dataset is carefully picked up where most of its instances relative area is less than 1% of the input image. The proposed network achieves 89.39%, which outperforms other state-of-the-art detectors.
It is of great practical significance to detect foreign object debris (FOD) timely and accurately on the airfield pavement, because the FOD is a fatal threaten for runway safety in airport. In this paper, a new FOD detection framework based on Single Shot MultiBox Detector (SSD) is proposed. Two strategies include making the detection network lighter and using dilated convolution, which are proposed to better solve the FOD detection problem. The advantages mainly include: (i) the network structure becomes lighter to speed up detection task and enhance detection accuracy; (ii) dilated convolution is applied in network structure to handle smaller FOD. Thus, we get a faster and more accurate detection system.
This paper proposed a novel region based multi-channel convolution neural network architecture for crowd counting. In order to effectively solve the perspective distortion in crowd datasets with a great diversity of scales, this work combines the main channel and three branch channels. These channels extract both the global and region features. And the results are used to estimate density map. Moreover, kernels with ladder-shaped sizes are designed across all the branch channels, which generate adaptive region features. Also, branch channels use relatively deep and shallow network to achieve more accurate detector. By using these strategies, the proposed architecture achieves state-of-the-art performance on ShanghaiTech datasets and competitive performance on UCF_CC_50 datasets.
The rapid development of electronic industry not only increases the variety of electronic components, but also increases the difficulty of inspection work. To effectively improve the appearance inspection performance, an appearance imaging system based on machine vision is proposed in this paper. The system provides practical solutions for the following four problems. Firstly, to maintain the consistence of appearance images, a standardized imaging method is presented to unify imaging parameters. Secondly, when dealing with different reflection properties, we proposed a combined illumination method using light sources with different wave lengths to meet imaging needs. Thirdly, for large size objects, we put forward a method combined with size measurement and image mosaic to get high-resolution and panoramic images. Fourthly, this paper provides a method of 3D reconstruction based on monocular vision to reflect depth information and geometric characteristics of real scenes. In the meanwhile, we design associated software to achieve the auto-control. Experiment results conclude that it is effective to achieve standardized and high-resolution appearance imaging. The system builds up a good foundation for the subsequent inspection work. The research of this paper has a broad meaning and application prospect.
KEYWORDS: 3D modeling, Cameras, Photography, Visual process modeling, Wind measurement, 3D acquisition, 3D metrology, 3D image processing, Imaging systems, Image processing
Schlieren photography is normal device in wind tunnel. It records varying density of flow and also shows the attitude of model. In this paper, a method is proposed to estimate the model attitudes through matching the projection drawings of 3D digital model with the schlieren photography and high speed camera image. A simulation experiment is also designed to test the method. The results show that the maximum error less than 0.1°. We also use the method to deal with the wind tunnel test data, and experimental results show that the proposed system can meet the demands of the wind tunnel test.
In this paper, we describe a fully automatic approach for detecting and matching geometrical corner feature correspondences between aerial images with larger scale and view variations. The main assumption of the approach is the fact that many man-made environments contain a large number of parallel linear features. We exploit this observation towards efficient detection and estimation of vanishing points. Given the vanishing points within an image, building geometrical corner features are obtained by the intersections of pairs of building outlines corresponding to different vanishing points. The experiments performed on the infrared aerial image sequences evaluate the stability and distinctiveness of the proposed features which are undergone appearance changes due to projective deformation.
KEYWORDS: 3D modeling, Cameras, Visual process modeling, Wind measurement, Imaging systems, High speed cameras, 3D image processing, Image processing, 3D metrology, Aerodynamics
A position and attitude vision measurement system for drop test slender model in wind tunnel is designed and developed.
The system used two high speed cameras, one is put to the side of the model and another is put to the position where the
camera can look up the model. Simple symbols are set on the model. The main idea of the system is based on image
matching technique between the 3D-digital model projection image and the image captured by the camera. At first, we
evaluate the pitch angles, the roll angles and the position of the centroid of a model through recognizing symbols in the
images captured by the side camera. And then, based on the evaluated attitude info, giving a series of yaw angles, a series
of projection images of the 3D-digital model are obtained. Finally, these projection images are matched with the image
which captured by the looking up camera, and the best match’s projection images corresponds to the yaw angle is the
very yaw angle of the model. Simulation experiments are conducted and the results show that the maximal error of
attitude measurement is less than 0.05°, which can meet the demand of test in wind tunnel.
Reliable modeling for thermal infrared (IR) signatures of real-world city scenes is required for signature management of civil and military platforms. Traditional modeling methods generally assume that scene objects are individual entities during the physical processes occurring in infrared range. However, in reality, the physical scene involves convective and conductive interactions between objects as well as the radiations interactions between objects. A method based on radiosity model describes these complex effects. It has been developed to enable an accurate simulation for the radiance distribution of the city scenes. Firstly, the physical processes affecting the IR characteristic of city scenes were described. Secondly, heat balance equations were formed on the basis of combining the atmospheric conditions, shadow maps and the geometry of scene. Finally, finite difference method was used to calculate the kinetic temperature of object surface. A radiosity model was introduced to describe the scattering effect of radiation between surface elements in the scene. By the synthesis of objects radiance distribution in infrared range, we could obtain the IR characteristic of scene. Real infrared images and model predictions were shown and compared. The results demonstrate that this method can realistically simulate the IR characteristic of city scenes. It effectively displays the infrared shadow effects and the radiation interactions between objects in city scenes.
Number extraction in document image is a crucial technique in document image analysis. To simply and efficiently
extract the numbers, a mathematical morphology based algorithm is proposed in this paper. Firstly, the square regions
containing numbers are labeled by morphological dilation operation using a designed structuring element. Secondly, the
square regions are extracted by morphological erosion operation. Thirdly, the inner region of the square regions is
extracted through morphological dilation operations. Finally, the numbers are extracted through comparing the extracted
and original inner region of the square regions. Experimental results show that the proposed algorithm is efficient.
A new infrared target segmentation algorithm by using watershed transform based on multi-scale mathematical
morphology and target enhancement is proposed in this paper. Firstly, the multi-scale mathematical morphological
operator is used to pre-process the original infrared image, which suppresses the effect of noises and protects targets.
Secondly, the property of the infrared image, non-parameter kernel method and linear extension are used to enhance dim
target. Thirdly, some pixels of the enhanced target regions are binarized and then processed by morphological operators
as the markers of the infrared targets. Finally, after the gradient of the pre-processed infrared image is calculated by
using Sobel detector, the watershed is performed on the gradient image guided by the markers of target regions to
segment the target regions. The proposed method can be widely used in different applications of target detection, target
tracking, navigation system and so on. Experimental results verify that the proposed method is efficient.
A novel marker based watershed through image enhancement is proposed to segment the dim infrared target. The dim
infrared target is firstly enhanced by CB top-hat transformation and image quantization. Then, the accurate marker of the
target can be easily obtained through image binarisation and marker filtering. To calculate an efficient gradient image of
the dim target for the watershed segmentation, the gradient image is firstly calculated through Sobel operator and then
efficiently enhanced through pseudo top-hat transformation and gradient quantization. Because of the enhancement of
the dim target and the gradient image, the watershed can efficiently segment the dim infrared image. Experimental
results show that the proposed algorithm is much efficient for dim infrared target segmentation.
To improve the performance of a top-hat transformation for infrared small target enhancement, a new class of top-hat transformation through structuring element construction and operation reorganization is proposed. The structuring element construction and operation reorganization are based on the property of the infrared small target image and thus can greatly improve the performance of small target enhancement. Experimental results verified that it was very efficient.
To reduce the influences of the dim target intensity and heavy clutter on infrared small target detection and tracking, a
novel algorithm is presented in this paper. The algorithm proposes a modified top-hat transformation by importing the
property of the small target region firstly, which largely enhances the dim target and apparently suppresses the heavy
clutter. Consequently, the potential targets are easy to be segmented by the iterative thresholding method. After
decreasing the false alarms through the dilation cumulation, the real target and the trajectory are correctly given by using
the data association formed by the motion property of the real target. Various experiments verified that the proposed
algorithm was efficient and robust for dim target detection and tracking under the condition of heavy clutter.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.