Faculty Publications
Permanent URI for this communityhttps://idr.nitk.ac.in/handle/123456789/18736
Publications by NITK Faculty
Browse
13 results
Search Results
Item Adaptive Learning Rate for Visual Tracking Using Correlation Filters(Elsevier B.V., 2016) Asha, C.S.; Narasimhadhan, A.V.Visual tracking is a difficult problem in computer vision due to illumination, pose, scale, appearance variations of object. Most of the trackers use either gray scale/color information or gradient information for image description. However the use of multiple channel features provide more information than single feature alone. Recently correlation filter based video tracking gained popularity due to its efficiency and high frame rate. Existing correlation filters use fixed learning rate to update filter template in every frame. In this paper, a method for adapting learning rate in correlation filter (CF) is presented which depends on the position of target in the present and previous frames (target velocity). This method uses integral channel features in correlation filter framework with adaptive learning rate to efficiently track the object. We experiment this technique on 12 challenging video sequences from visual object tracking (VOT challenges) datasets. Proposed technique can track any object irrespective of illumination variance, occlusion, scale change and outperforms the state-of-the-art trackers. © 2016 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license.Item Assessment of speckle denoising in ultrasound carotid images using least square Bayesian estimation approach(Institute of Electrical and Electronics Engineers Inc., 2017) Yamanakkanavar, Y.; Asha, C.S.; Narasimhadhan, A.V.The ultrasound carotid images affected by speckle noise, which highly reduces the image quality and effects the human interpretation. Speckle removal is substantial and critical step for preprocessing of ultrasound carotid images. For robust diagnosis, the carotid images must be free of noise and clear in clinical practices. The carotid ultrasound images have multiplicative noise and is very difficult to remove as compared to additive noise. To address this issue we propose to use Bayesian least square estimation in the logarithmic space. The proposed algorithm is tested on 50 ultrasound B mode carotid images and the performance of the algorithm is compared with the existing algorithms like Median filter, Speckle Reducing Anisotropic Diffusion(SRAD), Non Local Mean (NLM) filter, Total Variation (TV), Detail Preserving Anisotropic Diffusion(DPAD) filter, Lee filter, Frost filter and Wavelet filter. Experimental result shows that proposed algorithm capable of achieving better results as compared to the other methods in terms of signal to noise ratio (SNR), peak signal to noise ratio (PSNR), Correlation of Coefficient (CoC), Structural Similarity Index Map (SSIM) and Image Quality Index(IQI) measures. As per visual inspection concerned the proposed approach is more effective in terms of suppression of noise and image enhancement. © 2016 IEEE.Item Thermal vision human classification and localization using bag of visual word(Institute of Electrical and Electronics Engineers Inc., 2017) Malpani, S.; Asha, C.S.; Narasimhadhan, A.V.Human detection in thermal images has recently gained a lot of attention in computer vision due to its large number of applications. The characteristics of thermal images are poor illumination, low contrast due to capturing devices and poor environment conditions. Human classification and localization are being done using bag of visual word method. Bag of visual word method has been widely used for visible spectrum. In this work, we have extended it to thermal images. A new human detection scheme is present for thermal image using SURF features with Bag of Word. SURF has been compared with different binary feature descriptors. SURF feature descriptor outperforms BRISK and FREAK feature descriptors in terms of accuracy, F-score. © 2016 IEEE.Item Experimental evaluation of feature channels for object tracking in RGB and thermal imagery using correlation filter(Institute of Electrical and Electronics Engineers Inc., 2017) Asha, C.S.; Narasimhadhan, A.V.Correlation filter based trackers are well studied for object tracking and shown great interest to the research community in recent years. The vast majority of the works make utilization of either color feature channels or Histogram of Gradient feature channels for object tracking in visual spectrum. However, the strength of feature channels varies from RGB videos to thermal infrared videos. Subsequently, an assessment of feature channels in RGB and thermal imagery is needed to select the best features. In this work, we study the performance of various feature channels under kernelized correlation filter framework in RGB recordings, by taking 33 videos from object tracking benchmark (OTB) dataset and thermal infrared recordings, by taking 25 thermal videos from Thermal InfraRed (LTIR) dataset. Performance of each feature channels in both imaging modes are quantified using distance precision score, overlap score, average center location error and speed metrics. The best performance is obtained when HOG and color name features are utilized for RGB videos and gradient and gabor features are used in thermal videos among selected feature sets in kernelized correlation filter framework. © 2017 IEEE.Item Vehicle Counting for Traffic Management System using YOLO and Correlation Filter(Institute of Electrical and Electronics Engineers Inc., 2018) Asha, C.S.; Narasimhadhan, A.V.Vehicle counting is a process to estimate the road traffic density to assess the traffic conditions for intelligent transportation systems. With the extensive utilization of cameras in urban transport systems, the surveillance video has become a central data source. Also, real-time traffic management system has become popular recently due to the availability of handheld/mobile cameras and big-data analysis. In this work, we propose video-based vehicle counting method in a highway traffic video captured using handheld cameras. The processing of a video is achieved in three stages such as object detection by means of YOLO (You Only Look Once), tracking with correlation filter, and counting. YOLO attained remarkable outcome in the object detection area, and correlation filters achieved greater accuracy and competitive speed in tracking. Thus, we build multiple object tracking with correlation filters using the bounding boxes generated by the YOLO framework. Experimental analysis using real video sequences shows that the proposed method can detect, track and count the vehicles accurately. © 2018 IEEE.Item Automatic Segmentation of Intima Media Complex in Common Carotid Artery using Adaptive Wind Driven Optimization(Institute of Electrical and Electronics Engineers Inc., 2019) Madipalli, P.; Kotta, S.; Dadi, H.; Yamanakkanavar, Y.; Asha, C.S.; Narasimhadhan, A.V.Cardiovascular diseases have been one of the leading causes of death and have been increasing in much of the developing world. Atherosclerosis, the accumulation of plaque on artery walls is the major for cardiovascular diseases. This is diagnosed by measuring the thickness of IMC of common carotid artery (CCA) in ultrasound images. In this paper, we present a completely automatic technique for segmentation of IMC in ultrasound images of CCA. The image is segmented using adaptive wind driven optimization (AWDO) technique. The denoising filter based on Bayesian least square approach and a robust enhancement technique is used in the pre-processing stage. The proposed method is evaluated on 60 ultrasound images and is compared with the state-of-The-Art methods. The experimental results show that the proposed method yields better results as compared to other methods. © 2018 IEEE.Item A Comparative Study of Illumination Invariant Techniques in Video Tracking Perspective(Taylor and Francis Ltd. michael.wagreich@univie.ac.at, 2020) Asha, C.S.; Narasimhadhan, A.V.Object tracking is being utilized in the field of computer vision over decades for video surveillance, human–computer interaction and robotic applications. Even though the state-of-the-art tracking technology is rapidly growing, few issues are still challenging such as illumination variation, pose variation, scale changes, occlusion, etc. Among these challenges, sudden illumination variation is more complicated which is not solved completely. Most of the current trackers, indeed work under controlled illumination conditions in outdoor and indoor environments. In this work, we study the effect of adding the photometric normalization techniques prior to tracking in order to minimize the drift during abrupt light changes of the median flow tracker (MFT). The tracker under investigation is based on the optical flow method and achieved remarkable results in the tracking literature. However, it drifts off during sudden illumination variation. To resolve this problem, pre-processing technique is incorporated just before tracking. Hence, we present an experimental study of various pre-processing techniques to improve the accuracy of the MFT. A total of eight state-of-the-art normalization techniques are summarized and tested in video tracking perspective. The experiments are carried out with the video sequences obtained from the object tracking benchmark dataset posing sudden illumination change as a challenge to analyze the modified tracker. A comparative analysis indicates that the modified tracker outperforms the baseline tracker in terms of precision score and overlap score. © 2019 IETE.Item Robust infrared target tracking using discriminative and generative approaches(Elsevier B.V., 2017) Asha, C.S.; Narasimhadhan, A.V.The process of designing an efficient tracker for thermal infrared imagery is one of the most challenging tasks in computer vision. Although a lot of advancement has been achieved in RGB videos over the decades, textureless and colorless properties of objects in thermal imagery pose hard constraints in the design of an efficient tracker. Tracking of an object using a single feature or a technique often fails to achieve greater accuracy. Here, we propose an effective method to track an object in infrared imagery based on a combination of discriminative and generative approaches. The discriminative technique makes use of two complementary methods such as kernelized correlation filter with spatial feature and AdaBoost classifier with pixel intesity features to operate in parallel. After obtaining optimized locations through discriminative approaches, the generative technique is applied to determine the best target location using a linear search method. Unlike the baseline algorithms, the proposed method estimates the scale of the target by Lucas-Kanade homography estimation. To evaluate the proposed method, extensive experiments are conducted on 17 challenging infrared image sequences obtained from LTIR dataset and a significant improvement of mean distance precision and mean overlap precision is accomplished as compared with the existing trackers. Further, a quantitative and qualitative assessment of the proposed approach with the state-of-the-art trackers is illustrated to clearly demonstrate an overall increase in performance. © 2017 Elsevier B.V.Item Carotid wall segmentation in longitudinal ultrasound images using structured random forest(Elsevier Ltd, 2018) Yamanakkanavar, Y.; Asha, C.S.; Teja A, H.S.; Narasimhadhan, A.V.Edge detection is a primary image processing technique used for object detection, data extraction, and image segmentation. Recently, edge-based segmentation using structured classifiers has been receiving increasing attention. The intima media thickness (IMT) of the common carotid artery is mainly used as a primitive indicator for the development of cardiovascular disease. For efficient measurement of the IMT, we propose a fast edge-detection technique based on a structured random forest classifier. The accuracy of IMT measurement is degraded owing to the speckle noise found in carotid ultrasound images. To address this issue, we propose the use of a state-of-the-art denoising method to reduce the speckle noise, followed by an enhancement technique to increase the contrast. Furthermore, we present a novel approach for an automatic region of interest extraction in which a pre-trained structured random forest classifier algorithm is applied for quantifying the IMT. The proposed method exhibits IMTmean ± standard deviation of 0.66mm ± 0.14, which is closer to the ground truth value 0.67mm ± 0.15 as compared to the state-of-the-art techniques. © 2018 Elsevier LtdItem Multi-Modal Medical Image Fusion with Adaptive Weighted Combination of NSST Bands Using Chaotic Grey Wolf Optimization(Institute of Electrical and Electronics Engineers Inc., 2019) Asha, C.S.; Lal, S.; Gurupur, V.P.; Saxena, P.U.P.Recently, medical image fusion has emerged as an impressive technique in merging the medical images of different modalities. Certainly, the fused image assists the physician in disease diagnosis for effective treatment planning. The fusion process combines multi-modal images to incur a single image with excellent quality, retaining the information of original images. This paper proposes a multi-modal medical image fusion through a weighted blending of high-frequency subbands of nonsubsampled shearlet transform (NSST) domain via chaotic grey wolf optimization algorithm. As an initial step, the NSST is applied on source images to decompose into the multi-scale and multi-directional components. The low-frequency bands are fused based on a simple max rule to sustain the energy of an individual. The texture details of input images are preserved by an adaptively weighted combination of high-frequency images using a recent chaotic grey wolf optimization algorithm to minimize the distance between the fused image and source images. The entire process emphasizes on retaining the energy of the low-frequency band and the transferring of texture features from source images to the fused image. Finally, the fused image is formed using inverse NSST of merged low and high-frequency bands. The experiments are carried out on eight different disease datasets obtained from Brain Atlas, which consists of MR-T1 and MR-T2, MR and SPECT, MR and PET, and MR and CT. The effectiveness of the proposed method is validated using more than 100 pairs of images based on the subjective and objective quality assessment. The experimental results confirm that the proposed method performs better in contrast with the current state-of-the-art image fusion techniques in terms of entropy, VIFF, and FMI. Hence, the proposed method will be helpful for disease diagnosis, medical treatment planning, and surgical procedure. © 2013 IEEE.