Object detection from hyperspectral images (HSIs) is an important issue but encounters a critical challenge that results in poor detection due to the variation of the detection object spectrum. Especially when the detection object area is large and widely distributed in HSIs, such spectral variability becomes more serious. Spectral variability can make false detection and leak detection in object detection very serious. The constrained energy minimization (CEM) algorithm is a classical object detection algorithm that only needs the object prior spectrum to achieve object detection, but the spectral variability will have a detrimental effect on the detection results of the CEM algorithm. To address the above problems, we propose a multiobject subspace projection sample weighted CEM (MSPSW-CEM) algorithm. The proposed method has the following capabilities: (1) it constructs object subspaces and detectors using multiple prior spectra of the detection object under spectral variability conditions and (2) it utilizes the subspace projection theory to weight the pixel spectra, so that the detector can better suppress the background information and highlight the object information. Extensive experiments were carried out on two sets of real-world HSIs, and it was found that MSPSW-CEM generally showed a better detection performance than other object detection methods.
According to the feature of strong correlation of remote sensing image, a target recognition method based on Constrained
Independent Component Analysis (CICA) via Compressed Sensing is put forward to realize the goal of remote sensing
image recognition. By using abundance nonnegative restriction and the abundance sum-to-one constraint, an Adaptive
Abundance Modeling (AAM) algorithm is proposed to ensure the reliability of the objective function. Then the CS
feature space classifier based on Constrained Independent Component Analysis of sparse signal is established, so as to
achieve recognition quickly. Experimental results show that the proposed algorithm can obtain more accurate results as
high as 90%, and improve the timeliness effectively.
KEYWORDS: Clouds, Reconstruction algorithms, 3D modeling, LIDAR, 3D acquisition, 3D scanning, 3D image processing, Data modeling, Data acquisition, Detection and tracking algorithms
As for the characteristic of the data acquired by laser radar and the three dimentional point cloud in disorder, and by
combining the abundant in three dimentional information of point cloud with the specific textural information of distance
images, we raised a new algorithm on the reconstruction of laser radar based on simplified point cloud and distance
images. In this article, we take advantage of the feature that Delaunay triangulation have to raise a simplified algorithm
to achieve the model network. In this algorithm, at first we build up the Delaunay triangulation, then comfirm the vector
by calculating the distance that every vertex in the network from the adjacency vertex, and then calculate the intersection
angle that the vector with triangle around; at the same time set the angular threshold in order to generate the new
Delaunay triangulation. Experimental results show that this algorithm can accomplish the simplication of triangulation
without affecting the accuracy of the modeling, along with the detailed, textural and shading information, we can achieve
3D reconstruction of the target images effectively.
Image corner detection is one of fundamental problems of computer vision and image processing research. Most of the corner detectors detect too much redundant information and increase unnecessary computational costs. We present a new mask that uses only the margin pixels of two circles. To avoid the unnecessary computing, when the mask is scanning areas of similar intensity, the mask will go on with the next nucleus directly. Then a fast corner detection algorithm based on a double circle mask is proposed. We test the performance of our algorithm on digital signal processing and the experimental results show that our new detector performs better and costs less time than the three other methods to which it is compared.
As the important reconnaissance and offensive weapon in future battlefield, Micro Aerial Vehicle (MAV) is applied
more and more widely in civil and military field. In the sea battlefield, ship classification applied to MAV could
effectively realize signals collection, force protection and strike to ship targets. At present, methods of ship classification
are mostly based on signals from radar, infrared or ultrasonic. However, because of large volume and complex
equipments, these methods can't meet the requirement of MAV. Thus, ship classification based on visible sensor is
chosen and it could solve volume and weight limits of MAV. In order to realize ship classification in MAV, ship
classification based on aerial images is first proposed and an effective robust algorithm for classification based on
modified Zernike moment invariants is proposed in this paper. The task of classification is that the ships are classified
into two categories, aircraft carrier and chaser. The experimental results show that the correct classification rate is more
than 92% and the algorithm proposed is effective to solve classification problem for ship targets in MAV.
According to analysis of runways geometric features in remote sensing images, a new airfield detection method
combining color, texture segmentation and shape analysis is presented, where the color and texture features are used for
global classification while shape information is used for local analysis. In order to extract airfield runway information, an
improved method based on direction and length filter is proposed, in which the useless scanning can be stopped promptly.
The experimental results presented in this paper show that this algorithm could eliminate the interference under complex
circumstance and could improve the efficiency and accuracy of military airfield recognizing and understanding. It has
higher computing speed and less space demand compared with the existing Hough-based algorithm. Furthermore, the
proposed algorithm is simple and easy to implement.
KEYWORDS: Magnetism, Sensors, Magnetic sensors, Digital signal processing, Amplifiers, Power supplies, Resistance, Mathematical modeling, Imaging systems, Imaging arrays
Pipeline transportation, as it is low cost, steady supply and high efficiency, is widely used nowadays. However, pipelines
might be damaged by natural power or human activities. Thus, pipeline status monitoring, including transmogrification,
corrosion, flaw and crack, shows up more and more important. This paper presents a method using magnetic field
measurement system which based on AMR (anisotropy magnetic resistance) sensors array to imaging the pipe's bug.
Compared with single sensor, it can capture more all-around information about the magnetic field distribution on pipe
wall, and it can make the detection more veracity; Compared with the traditional pipeline pig, which based on Hall
elements, it provides greater sensitivity. A mechanical model of relationships between the pipe's bug and the magnetic
field distribution is given; In this AMR measurement system, the hardware includes arrangement sensors, Set/Reset
circuits, amplifiers, multiplexers, DSP device, data radio module and power supply; and software contains an autocalibration
algorithm, and a VC display program of the measured magnetic field. An experimental pipe bug is detected
and the magnetic field is discussed.
KEYWORDS: Digital signal processing, Transducers, Resistance, Signal processing, Neural networks, Complex systems, Digital electronics, Neurons, Analog electronics, Evolutionary algorithms
For the purpose of better application, the nonlinear correction of the transducer is very important. In this
paper, the nonlinear correction system for the thermal resistance transducer is researched to realize the
nonlinear correction. The system consists of five parts including the thermal resistance transducer, the
amplifying circuit, AD converting circuit, digital signal processor (DSP) unit and display module. As for the
algorithms that are applied in the system, the neural network method and linear interpolation method are
discussed. And nonlinear correction experiments by means of the two kinds of algorithms were done. Through
the experiments, it is proved that the nonlinear correction system based on DSP is able to realize the nonlinear correction
of the thermal resistance transducer, and the neural networks algorithm is more effective and accurate than the linear interpolation method.
The paper presents a application method of detecting moving ground target based on micro accelerometer. Because vehicles moving over ground generate a succession of impacts, the soil disturbances propagate away from the source as seismic waves. Thus, we can detect moving ground vehicles by means of detecting seismic signals using a seismic tranasducer, and automatically classify and recognize them by data fusion method. The detection system on the basis of MEMS technology is small volume, light weight, low poer, low cost and can work under poor circumstances. In order to recognize vehicle targets, seismic properties of typical vehicle targets are researched in the paper. A data fusion technique of artifical neural networks (NAA) is applied to recognition of seismic signals for vehicle targets. An improved back propagation (BP) algorithm and ANN architecture have been presented to improve learning speed. The improved BP algorithm had been used recognition of vehicle targets in the outdoor environment. Through experiments, it can be proven that target seismic properties acquired are correct. ANN data fusion is effective to solve the recognition problem of moving vehicle target, and the micro accelerometer can be used in target recognition.
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.