Mahboob Iqbal

Scene classification for aerial images based on CNN using sparse coding technique

ABSTRACT Aerial scene classification purposes to automatically label aerial images with specific semantic categories. However, cataloguing presents a fundamental problem for high-resolution remote-sensing imagery (HRRS). Recent developments include several approaches and numerous algorithms address the task. This article proposes a convolutional neural network (CNN) approach that utilizes sparse coding for scene classification applicable for HRRS unmanned aerial vehicle (UAV) and satellite imagery. The article has two major sections: the first describes the extraction of dense multiscale features (multiple scales) from the last convolutional layer of a pre-trained CNN models; the second describes the encoding of extracted features into global image features via sparse coding to achieve scene classification. The authors compared experimental outcomes with existing techniques such as Scale-Invariant Feature Transform and demonstrated that features from pre-trained CNNs generalized well with HRRS datasets and were more expressive than low- and mid-level features, exhibiting an overall 90.3% accuracy rate for scene classification compared to 85.4% achieved by SIFT with sparse coding. Thus, the proposed CNN-based sparse coding approach obtained a robust performance that holds promising potential for future applications in satellite and UAV imaging.

Disparity map estimation based on optimization algorithms using satellite stereo imagery

The monitoring of trees and vegetation near high voltage transmission power lines is a tedious job for electrical companies. There are many blackouts occur due to interfering the trees with the power transmission lines in hilly as well as urban areas. This is a big challenge for power distribution companies to monitor the vegetation for avoiding the blackouts and flashovers. To solve these problems, there are many methods are used to monitor the trees and vegetation near transmission power poles. But the existing methods are expensive and time consuming. We proposed the new method based on satellite images to monitor the trees and vegetation. The satellite images provide the cost effective solution to solve the monitoring problem. In this paper, we proposed the stereo matching algorithms to measure the disparity map based on satellite stereo imagery. The height estimation of trees and vegetation near power poles based on depth map which is inversely proportional of the disparity map. For measuring the depth map, the dynamic programming (DP) and block matching with energy minimization has been proposed. These techniques are applied on the satellite stereo images and based on results, our proposed DP algorithm produced more accurate disparity map as compared to the block matching algorithm.

Measuring height of high-voltage transmission poles using unmanned aerial vehicle (UAV) imagery

ABSTRACT Aerial imagery is important in remote sensing applications. Unmanned aerial vehicle (UAV) has a wide range of applications in remote sensing and presents a substantial cost-effective solution when monitoring objects on the earth’s surface. Moreover, object detection and classification are important aspects of global information system, especially for remote sensing applications and power line monitoring, which are essential for the proper distribution of electricity to consumers. Manual inspection consumes much time and involves risk, especially in remote areas that host dangerous wildlife; hence, UAV-based approaches are more feasible for such monitoring. The authors propose an UAV approach that utilises a digital surface model and incorporates a stereo matching algorithm based on UAV stereo images. The proposed algorithm was based on a graph-cut (GC) algorithm that measured the disparity map. Results were compared with well-known algorithms; including, for example, global and local stereo matching algorithms. The proposed solution introduces and integrates ordering constraints along with a submodular energy minimisation function to/with the GC algorithm to enhance performance. The authors measured sensitivity and recall for all parameters against ground truth data for differently cropped images of 16 power poles. Results showed that the proposed model performed more accurately compared to extant methods.

Image classification based on sparse-coded features using sparse coding technique for aerial imagery: a hybrid dictionary approach

This work offers an approach to aerial image classification for use in remote sensing object recognition, image processing and computer vision. Sparse coding (SC) is used to classify unmanned-aerial-vehicle (UAV) and satellite images because SC representation can generalize a large dataset and improve the detection of distinctive features by reducing calculation time for feature matching and classification. Features from images are extracted based on the following descriptors: (a) Scale Invariant Feature Transform; (b) Histogram of Oriented Gradients; and (c) Local Binary Patterns. SC representation and local image features are combined to represent global features for classification. Features are deployed in a sparse model to store descriptor features using extant dictionaries such as (a) the Discrete Cosine Transform and (b) the Discrete Wavelet Transform. An additional two dictionaries are proposed as developed for the present work: (c) the Discrete Ridgelet Transform (DRT) and (d) the Discrete Tchebichef Transform. The DRT dictionary is constructed by using the Ricker wavelet function to generate finite Ridgelet transforms as basis elements for a hybrid dictionary. Different pooling methods have also been employed to convert sparse-coded features into a feature matrix. Various machine learning algorithms are then applied to the feature matrix to classify objects contained in UAV and satellite imagery data. Experimental results show that the SC model secured better accuracy rates for extracted discriminative features contained in remote sensing images. The authors concluded that the proposed SC technique and proposed dictionaries provided feasible solutions for image classification and object recognition.

Design of digital elevation model based on orthorectified satellite stereo images

In this paper, orthorectification of satellite data from Pleiades satellite sensors based on ground control points (GCPs) is performed. A true orthorectification of satellite data is possible with an accurate digital elevation model (DEM). This study was challenging due to weather condition, hilly, and nonuniform terrain at the city Kota Kinabalu in East Malaysia. First, the geometric accuracy of the sensors is required for 3D mapping of the area of interest. In addition, an accurate DEM for better height calculation of vegetation/trees near power poles is required. For these purposes, first we need the true orthorectification of satellite stereo imagery. The measured GCP is compared with the actual GCPs to evaluate the geometric correction in terms of root mean square error. The comparison was made based on rational polynomial Function (RPC) coefficients and GCP measurements. Results show that the Pleiades satellite sensor is feasible for generating the elevation model for small area. This shows that Pleiades sensor can be used for monitoring the vegetation near poles of high transmission power lines based on the satellite stereo images.

Design of dictionary based on Discrete Tchebichef Transform

In this paper, design of overcomplete dictionaries based on Discrete Cosine Transform and Discrete Tchebichef Transform using optimization techniques is presented. Further, we optimized our proposed dictionaries based on KSVD algorithm and measured the performance of dictionaries using orthogonal matching pursuit (OMP) and basis pursuit (BP). The result showed that the Dictionary based on Discrete Tchebichef Transform (DTT) performed better as compared to the dictionary based on Discrete Cosine Transform (DCT). The proposed transform is first time introduced to made comparison with the DCT based dictionary generation. The proposed dictionaries are predetermined and optimize using KSVD algorithm. The accuracy will be increase with the slight increase of the computation complexity using Discrete Tchebichef Transform as compared to the Discrete Cosine Transform. The root mean square values are used to measure the accuracy.

Dynamic programming based comparison including QuickBird and IKONOS satellite stereo images for monitoring vegetation near power poles

Vegetation encroachment under overhead high voltage power line and its monitoring is a challenging problem for electricity distribution companies. Blackout can occurs if proper monitoring of vegetation is not done. The uninterrupted electric power supply is vital for industries, businesses, and daily life. Therefore, it is mandatory for electricity companies to monitor the vegetation/trees near power lines to avoid the blackouts. Many approaches are employed to monitor vegetation/trees near the transmission line poles, but these approaches are more time consuming and expensive. In this paper, we have proposed a novel approach to monitor the vegetation near or under the power poles using satellite stereo images which were acquired using QuickBird and IKONOS satellites. 3D depth of vegetation has been measured using stereo algorithm incorporating dynamic programming. We have also compared the results of QuickBird and IKONOS satellite stereo images. Results showed that QuickBird satellite image performs well as compared to IKONOS using stereo vision global optimization dynamic programming algorithm in terms of accuracy and speed.

Monitoring of vegetation near power lines based on dynamic programming using satellite stereo images

Vegetation encroachment of high voltage power line endorsement space is an exceeding problem for electricity distribution companies. The electrical utilities have responsibility to impose their vegetation management proceeding so as to evade vegetation/ trees near transmission power lines. When the height of trees/vegetation is increased and it makes a contact with the power lines, it may antecedent the power lines in result of blackouts. Blackouts come about due to vegetation encroachments can cause impressive compensation. The uninterrupted electric supply is very imperative for industries, businesses, and populous areas. It is indispensable for electricity companies to monitor the vegetation/trees near power lines, There are many approaches applicable to monitor vegetation/trees near the transmission line poles, but these approaches are time inefficient in terms of time and finance. In this paper, a novel technique for depth estimation of vegetation/trees is proposed. In the study, Dynamic Programming is employed on stereo satellite images to determine depth of vegetation and trees. The experimental results on QuickBird imagery exhibit that the proposed technique performs better compared to block matching technique in terms of accuracy.

Power LinesVegetation enchroachment monitoring based on Satellite Stereo images using stereo matching

Continuous monitoring the vegetation encroachment for high voltage transmission lines is challenging task for electrical distribution authorities. The enchroachment causes interruption which results in blackouts. It also endures great cost to the authorities for maintenance and damage compensation. There are many methods available to monitor growth of vegetation near transmission line poles. However, these methods are constrained with time consumption and cost. In this paper,a new method based estimating 3D height/depth of vegetation or trees near transmission lines using satellite stereo vision is proposed. We have evaluated local methods of stereo matching using Quickbird Satellite stereo images and compare the performance of stereo matching cost functions in terms of processing speed and time. Results show that our proposed algorithm is faster and more accurate than the existing algorithm for disparity calculation.