Target recognition by texture segmentation algorithm

This paper presents a novel algorithm for unsupervised texture segmentation. The proposed algorithm incorporates the local binary pattern operator under a segmentation framework based on the active contour without edges model. The experiments performed, show that it can be used for fast segmentation of two-textured images, outperforming recent texture segmentation algorithms, with a segmentation quality that reaches 99% on average

Motivated by the self-assembling behavior of real ants, we present a novel algorithm for texture segmentation which is based on ant tree clustering of wavelet features. In a pattern recognition setting, wavelet features are extracted using either of the two subband filtering methods: discrete wavelet transform (DWT) or discrete wavelet packet transform (DWPT). The feature classification process is inspired by the self-assembling behavior observed in real ants where ants progressively become attached to an existing support and then successively to other attached ants thus building trees based on the similarity of feature vectors. The results thus obtained compare favorably to those of other recently published filtering based texture segmentation algorithms

The purpose of the research is to develop a universal algorithm for partial texture segmentation of any visual images. The main peculiarity of the proposed segmentation procedure is the extraction of only homogeneous fine-grained texture segments present in the images. At first, an initial seed point is found for the largest and most homogeneous segment of the image. This initial seed point of the segment is expanded using a region growing method. Other texture segments of the image are extracted analogously in turn. At the second stage, the procedure of merging the extracted segments belonging to the same texture class is performed. Then, the detected texture segments are input to a neural network with competitive layers which accomplishes more accurate delineation of the shapes of the extracted texture segments. The proposed segmentation procedure is fully unsupervised, i.e., it does not use any a priori knowledge on either the type of textures or the number of texture segments in the image. The research results in development of the segmentation algorithm realized as a computer program tested in a series of experiments that demonstrate its efficiency on grayscale natural scenes.

In the study of object recognition, image texture segmentation has being a hot and difficult aspect in computer vision. Feature extraction and texture segmentation algorithm are two key steps in texture segmentation. An effective texture description is the important factor of texture segmentation. In this paper, the neuronal activation degree (NAD) of visual model is exploited as the texture description of image patches. By processing the length and direction of NAD, we develop an effective segmentation strategy. First, the length of the NAD are used to partition blank area and non-blank area, then the mark index of neuron is used, which is maximally activated to identify the label of each segment unit to get an initial segmentation. Finally, region merging steps is exerted to get a desired result.

Relevant with some important subjects like target recognition, sensor fusion systems can be considered as one of the main issues highlighting here. Environmental condition, target characteristic and sensor efficiency are three parameters which can impress on sensor value in target recognition so for recognizing targets, a group of sensors which have more recognition rates, must be selected intelligently. Utilizing many sensors to acquire the highest object recognition rate would have extra cost and decrease energy of mobile sensors rapidly. Therefore make a tradeoff between sensor numbers and object recognition rate would be imperatively. This paper attempts to design a multi objective optimization service by using optimization algorithm and neural network. This service specifies highest recognition rate for each distinct sensor numbers. We propose multi objective optimization algorithm to help accessing the best sensory configuration for a definite environment regarding to the environmental conditions, sensors performance, and object features. Our multi objective optimization algorithm has two functions. Genetic algorithm is used to perform as one of functions to specify object recognition rates of each sensor group. Neural network is used to perform as fitness function of each genetic algorithm chromosomes. Another function is sensor numbers determinant. Highest recognition rate and lowest sensor numbers are two objects which multi objective optimization algorithm wants to make a balance between them. We define 500 different scenarios for 6 different sensors in different conditions. Object recognition rate of each sensor is collected. These rates are used for neural networks training process. By defining new scenario and run multi objective optimization algorithm in this scenario, this algorithm makes a Pareto front between sensor numbers and object recognition rate. Finally this algorithm finds, by distinct numbers of sensor, which sensors by which recognition ability must be used to reach the highest recognition rate.

From the computer science literature, it can be seen that many different technologies are used in target recognition, which is one of the most significant areas in the artificial intelligence field. Target recognition is applied in a variety of disciplines, including healthcare, robot vision, vehicular traffic, and virtual reality. Target recognition techniques involve a robotic vision system that must perform with high accuracy and efficiency in real time; additionally, it must have the capacity to handle difficult identification contexts. In one existing target recognition system, the Harris algorithm is used; it provides a higher accuracy compared to more traditional algorithms. In order to improve its achieved accuracy, we focus on the target detection algorithm of a rehabilitation robot that is based on the local features of images. Considering the feature points of the images and target identification technology, a rehabilitation robotic recognition method is developed in this work. Initially, it collects the images, and then, adaptive weighted symplectic geometry decomposition is used for pre-processing. This method helps to reduce the noise in the images. Next, the features are extracted, and the vectors of the features are separated and identified. Afterward, one-to-many rehabilitation modes and actual system monitors are implemented to precisely select the target condition based on the functional criteria of the rehabilitation robot recognition method. Finally, an invertible color-to-grayscale conversion method using clustering and reversible watermarking is applied. It converts images into grayscale. The Gaussian distribution is consistently utilized to define the position and the quantity of the extracted feature points. Related images are retrieved as well. According to experimental findings, the proposed method improves the accuracy and the recall rate compared with the Harris algorithm.

Introduction. An algorithm for calculating the similarity degree between multidimensional histograms is presented. The proposed algorithm was intended for texture segmentation of images using histograms as texture features. The need to develop such a special algorithm is justified by the fact that the methods for estimating the similarity/difference measure between multidimensional vectors described in the literature provide such measures that are not very suitable for solving the texture segmentation task. The main peculiarity of the proposed algorithm is that when calculating the similarity value, it considers not only the corresponding histogram components, but also takes into account their nearest neighboring components. Due to this, the algorithm more adequately evaluates the similarity of histograms. The proposed algorithm was implemented as a computer program as an integral part of the image segmentation model. The effectiveness of the histogram comparison algorithm was indirectly confirmed by the results of texture segmentation of the image segmentation model in experiments on processing various images, including natural landscapes. Methods. The task of calculating the similarity between histograms is considered. A special algorithm is proposed because the analogical methods described in the literature are not very suitable for solving the texture segmentation task. The main peculiarity of the algorithm is that it takes into account as the corresponding histogram components as their nearest neighboring components. Due to this, the algorithm more adequately evaluates the similarity of histograms. The algorithm was implemented as a computer program. The effectiveness of the algorithm is indirectly confirmed by the results of texture segmentation of the image segmentation model in experiments on processing various images, including natural landscapes. Purpose. The goal of this work is to develop an efficient algorithm for assessing the similarity of histograms, such as brightness histograms and orientation histograms of the texture windows. The algorithm is based on the idea of taking into account not only the corresponding components of both histograms, but also the components of their immediate environment. Results. The main advantage of the proposed algorithm, compared to popular methods of calculating similarity/difference between objects (vectors), is that the range of similarity between the compared histograms (from complete similarity to complete difference) is 100%, while popular methods can offer several times smaller ranges of similarity percentage. Conclusion. The proposed algorithm provides a wide range of similarity between the compared histograms which is 100% (from complete similarity to complete difference), while popular methods can offer several times smaller ranges of similarity percentage. The algorithm was implemented as a computer program as a component of a model that solves the problem of segmenting a visual image into homogeneous texture areas. It is worth noting that the proposed histogram comparison algorithm calculates the similarity measure between histograms very quickly, since it uses only simple operations. The effectiveness of the algorithm for texture segmentation of images into homogeneous texture areas is confirmed by the results in the experiments on natural image processing. The results obtained in the experiments demonstrate the effectiveness of the algorithm and show that the algorithm performs correct (from a human point of view) texture segmentation of a wide range of images. Thus, the effectiveness of the key operation of the segmentation algorithm, the histogram comparison algorithm, is indirectly confirmed.

Cross-task and cross-domain SAR target recognition: A meta-transfer learning approach

Accurate localization of landmarks in the vicinity of a robot is a first step towards solving the SLAM problem. In this work, we propose algorithms to accurately estimate the 3D location of the landmarks from the robot only from a single image taken from its on board camera. Our approach differs from previous efforts in this domain in that it first reconstructs accurately the 3D environment from a single image, then it defines a coordinate system over the environment, and later it performs the desired localization with respect to this coordinate system using the environment's features. The ground plane from the given image is accurately estimated and this precedes segmentation of the image into ground and vertical regions. A Markov Random Field (MRF) based 3D reconstruction is performed to build an approximate depth map of the given image. This map is robust against texture variations due to shadows, terrain differences, etc. A texture segmentation algorithm is also applied to determine the ground plane accurately. Once the ground plane is estimated, we use the respective camera's intrinsic and extrinsic calibration information to calculate accurate 3D information about the features in the scene.

In this paper, a novel texture segmentation algorithm is proposed to partition the complex aerial insulator images into sub-regions with closed smooth contours. Firstly, Gray Level Co-occurrence Matrix (GLCM) is employed to extract the texture features of insulators and is calculated by the rapid Gray Level Co-occurrence Integrated Algorithm (GLCIA). We divide the extracted texture features into two categories: one with the stronger discriminative ability and the other with weaker ability. The second category is optimized by Principal Component Analysis (PCA) to better distinguish the different texture objects with low contrast. Then, a new convex energy functional is defined by taking the non-convex model of the Texture Descriptor Active Contour (TDAC) into a global minimization framework (GMAC) during segmentation. The proposed energy functional can avoid the existence of local minima in the minimization of the TDAC. A fast dual formulation is introduced for the efficient evolution of the contour. The experimental results on synthetic and real aerial insulator remote sensing images have shown that the proposed algorithm obtains more satisfactory segmentation compared to the classical models in terms of accuracy, efficiency and independence of initial contour. The influence of the algorithm parameters is also analyzed.

As the coverage area of the sub-block is too large, the texture feature clustering based on sub-block often produces the mosaic phenomenon of inaccurate boundary. In this paper, the texture clustering algorithm based on pixel extracts the texture feature vector of its central pixel point from sub-block. The image texture feature is standardized, then the normalized feature vector is clustered and the clustering result is used to realize the segmentation of complex image. Compared with the traditional segmentation methods such as gradient method, threshold method and so on, the segmentation boundary is accurate, and the phenomenon of under-segmentation and over-segmentation also reduce significantly.

The objective of this work is to simulate visual cortical cells, their sensitivities to frequencies and orientations, and their part in texture segmentation. The simulation of these cells is realised through band-pass, oriented filters (Gabor filters), and multiresolution image decomposition. By this means, the filter sensitivities represent cell sensitivities to preferred orientations according to their frequency and orientation bandwidths, and multiresolution represents the different band frequencies. For texture analysis and segmentation, overlaying of band-pass filters is necessary to completely cover the Fourier domain. A continuous sensitivity to frequency and orientation is achieved by the filters overlapping and consequently by their interpolation. We used here four octave frequency bands from 1 to 16 cycles deg−1 and six orientations per band. The results obtained for texture segmentation with these parameters are very promising (up to 97% recognition rate) [Guérin-Dugué and Palagi, 1994 Neural Processing Letters1(1) 25 – 29]. The images analysed cover a multitude of different domains such as psychophysical tests and natural textures of different roughness. In order to create a cortical cell representation closer to neurophysiological data, and to improve texture segmentation results, we represent cell sensitivities by their foveal and parafoveal projections [R L DeValois, K K DeValois, 1988 Spatial Vision (Oxford: Oxford Science Publications)]. Cells receiving projections from the foveal zone are modeled by five octave frequency bands (from 0.5 to 16 cycles deg−1) and six orientations. Cells receiving projections from the parafoveal zone have the same sensitivities but are modeled by four octave frequency bands (from 0.5 to 8 cycles deg−1). By using these two different resolutions, preliminary tests have shown the capability of detecting textured regions by the parafoveal projection and localisation of boundaries by the foveal projection.

Automatic target recognition (ATR) is a challenging task for several computer vision applications. It requires efficient, accurate, and robust methods for target detection and target identification. Deep learning has shown great success in many computer vision applications involving color RGB images. However, the performance of these networks in ATR with infrared sensor data needs further investigation. In this paper, we propose a multistage automatic target detection and recognition (ATDR) system that performs both target detection and target classification on infrared (IR) imagery using deep learning. Our system processes large IR image frames where targets take <1 % of the total number of pixels. First, we train a state-of-the-art object detector you only look once (YOLO) to localize all potential targets in the input image frame. Then, we train a convolutional neural network (CNN) to identify these detections as targets or false alarms. In this second phase, we adapt and analyze the performance of three CNN architectures: a compact and fully connected CNN, VGG16 with batch normalization, and a wide residual neural network (WRN). We also explore the use of a loss function that optimizes directly the area under the receiver operating characteristic (ROC) curve (AUC), and adapt it to our ATR application. To enhance the robustness of the proposed ATR to perturbation and variations introduced during the detection stage, we train our CNN classifiers on automatically detected targets using YOLO, in addition to ground truth bounding boxes and apply selected data augmentation techniques. To simulate real testing environments, where the spatial location of the targets within the image frame is unknown, only YOLO-detected boxes are used during validation. We evaluate our ATDR on a real benchmark dataset that includes different vehicles captured at different resolutions. Our experiments have shown that YOLO can detect most of the targets at the expense of generating a high number of false alarms. We show that the VGG-16 network with batch normalization, which is the best performing model, can correctly identify the classes of the targets, as well as classify the majority of YOLO’s false detections into an additional nontarget class. We also show that the proposed training modification to optimize an AUC-based loss function for ATR proved to be advantageous mainly in identifying difficult targets.

Automatic target recognition (ATR) is of increasing importance for the modern radar system, where the high-resolution range profile (HRRP) is essential. However, the recognition accuracy and sensitivity to dataset should be optimized for practical applications. Herein, we proposed a novel algorithm for HRRP target recognition based on a convolutional capsule network rather than neurons in conventional deep neural networks. The capsules were vectors trained to learn latent features from input HRRP, with the length of the vector representing the confidential probability. The convolution dynamic routing mechanism was applied between capsule layers by shared transformation matrices and constrained routing procedures in local kernels, reducing the size of parameters and computational expense. Experiments on measured data proved that the proposed algorithm outperforms other existing methods with higher recognition accuracy and less sensitivity to training size. This study provided a promising and effective approach for HRRP target recognition.

The synthetic aperture radar (SAR) image preprocessing techniques and their impact on target recognition performance are researched. The performance of SAR target recognition is improved by composing a variety of preprocessing techniques. The preprocessing techniques achieve the effects of suppressing background redundancy and enhancing target characteristics by processing the size and gray distribution of the original SAR image, thereby improving the subsequent target recognition performance. In this study, image cropping, target segmentation, and image enhancement algorithms are used to preprocess the original SAR image, and the target recognition performance is effectively improved by combining the above three preprocessing techniques. On the basis of image enhancement, the monogenic signal is used for feature extraction and then the sparse representation-based classification (SRC) is used to complete the decision. The experiments are conveyed on the moving and stationary target acquisition and recognition (MSTAR) dataset, and the results prove that the combination of multiple preprocessing techniques can effectively improve the SAR target recognition performance.