Serdar Cakir

Multiscale contrast direction adaptive image fusion technique for MWIR-LWIR image pairs and LWIR multifocus infrared images

Infrared (IR) cameras are widely used in the latest surveillance systems because spectral characteristics of objects provide valuable information for object detection and identification. To assist the surveillance system operator and automatic image processing tasks, fusing images in the IR band was performed as a solution to increase situational awareness and different fusion techniques were developed for this purpose. Proposed techniques are generally developed for specific scenarios because image content may vary dramatically depending on the spectral range, the optical properties of the cameras, the spectral characteristics of the scene, and the spatial resolution of the interested targets in the scene. In this study, a general purpose IR image fusion technique that is suitable for real-time applications is proposed. The proposed technique can support different scenarios by applying a multiscale detail detection and can be applied to images captured from different spectral regions of the spectrum by adaptively adjusting the contrast direction through cross-checking between the source images. The feasibility of the proposed algorithm is demonstrated on registered multispectral [mid-wave IR (MWIR), long-wave IR (LWIR)] and LWIR multifocus images. Fusion results are presented and the performance of the proposed technique is compared with the baseline fusion methods through objective and subjective tests. The technique outperforms baseline methods in the subjective tests and provide promising results in objective quality metrics with an acceptable computational load. In addition, the proposed technique preserves object details and prevents undesired artifacts better than the baseline techniques in the image fusion scenario that contains four source images.

Classifier-based offline feature selection and evaluation for visual tracking of sea-surface and aerial targets

An offline feature selection and evaluation mechanism is used in order to develop a robust visual tracking scheme for sea-surface and aerial targets. The covariance descriptors, known to constitute an effi- cient signature set in object detection and classification problems, are used in the feature extraction phase of the proposed scheme. The per- formance of feature sets are compared using support vector machines, and those resulting in the highest detection performance are used in the covariance based tracker. The tracking performance is evaluated in dif- ferent scenarios using different performance measures with respect to ground truth target positions. The proposed tracking scheme is observed to track sea-surface and aerial targets with plausible accuracies, and the results show that gradient-based features, together with the pixel locations and intensity values, provide robust target tracking in both surveillance scenarios. The performance of the proposed tracking strategy is also compared with some well-known trackers including correlation, Kanade- Lucas-Tomasi feature, and scale invariant feature transform-based track- ers. Experimental results and observations show that the proposed target tracking scheme outperforms other trackers in both air and sea surveil- lance scenarios. C

Image reconstruction for single detector rosette scanning systems based on compressive sensing theory

Abstract. Compressive sensing (CS) is a signal processing technique that enables a signal that has a sparse representation in a known basis to be reconstructed using measurements obtained below the Nyquist rate. Single detector image reconstruction applications using CS have been shown to give promising results. In this study, we investigate the application of CS theory to single detector infrared (IR) rosette scanning systems which suffer from low performance compared to costly focal plane array (FPA) detectors. The single detector pseudoimaging rosette scanning system scans the scene with a specific pattern and performs processing to estimate the target location without forming an image. In this context, this generation of scanning systems may be improved by utilizing the samples obtained by the rosette scanning pattern in conjunction with the CS framework. For this purpose, we consider surface-to-air engagement scenarios using IR images containing aerial targets and flares. The IR images have been reconstructed from samples obtained with the rosette scanning pattern and other baseline sampling strategies. It has been shown that the proposed scheme exhibits good reconstruction performance and a large size FPA imaging performance can be achieved using a single IR detector with a rosette scanning pattern.

Compressive sensing applications for single detector rosette scanning infrared seekers

Compressive sensing (CS) theory states that a signal which can be sparsely represented in a known basis may be reconstructed from its samples which have been obtained below the Nyquist rate. Image reconstruction with a single detector using CS theory has been shown to give promising results. In this work, we investigate the application of CS theory to single detector infrared (IR) rosette scanning systems. The single detector pseudo-imaging rosette scanning system scans the scene with a specific pattern and performs processing to estimate the target location without forming an image. These systems suffer from low performance compared to costly focal plane array (FPA) detectors. Using the CS framework, these scanning systems may be improved by reconstructing the samples obtained by the rosette scanning pattern. For this purpose, we consider surface to air engagement scenarios where the IR images contain aerial targets and flares. The IR images have been reconstructed from samples obtained with the rosette scanning pattern and other baseline sampling strategies. It has been shown that the proposed scheme exhibits good reconstruction performance and large size FPA imaging performance can be achieved using a single IR detector with a rosette scanning pattern.

Image generation for single detector infrared seekers via compressive sensing

In this paper, we investigate the application of compressive sensing theory to single detector infrared seekers. Compressive sensing is a novel signal processing technique which enables a compressible signal to be constructed using fewer measurements obtained in a specific way below the Nyquist rate. Single detector image reconstruction applications using compressive sensing have been shown to be successful. Infrared seekers utilizing single detectors suffer from low performance compared to costly focal plane array detectors. The single detector, pseudo-imaging rosette scanning seekers scan the scene with a specific pattern and process the resultant signal with signal processing methods to estimate the target location without forming an image. In this context, this type of old generation seekers can be converted to imaging systems by utilizing the samples obtained by the scanning pattern in conjunction with the compressive sensing theory framework. In this study, infrared images have been reconstructed from samples obtained by the rosette scanning pattern for different sample numbers and it has been shown that the results obtained are comparable to the results obtained by other sampling methods proposed in the literature.

Sampling strategy for the sparse recovery of infrared images

The compressive sensing (CS) framework states that a signal that has a sparse representation in a known basis may be reconstructed from samples obtained at a sub-Nyquist sampling rate. The Fourier domain is widely used in CS applications due to its inherent properties. Sparse signal recovery applications using a small number of Fourier transform coefficients have made solutions to large-scale data recovery problems, including image recovery problems, more practical. The sparse reconstruction of 2D images is performed using the sampling patterns generated by taking the general frequency characteristics of the images into account. In this work, instead of forming a general sampling pattern for infrared (IR) images, a special sampling pattern is obtained by gathering a database to extract the frequency characteristics of IR sea-surveillance images. Experimental results show that the proposed sampling pattern provides better sparse recovery results compared to the widely used patterns proposed in the literature. It is also shown that, together with a certain image dataset, the sampling pattern generated by the proposed scheme can be generalized for various image sparse recovery applications.

A multiscale contrast direction adaptation approach for the fusion of multispectral and multifocus infrared images

Infrared (IR) cameras are widely used in latest surveillance systems because spectral characteristics of objects provide valuable information for object detection and identification. To assist the surveillance system operator and automatic image processing tasks, fusing images in IR band is proposed as a solution to increase situational awareness and different fusion techniques are developed for this purpose. Proposed techniques are generally developed for specific scenarios because image content may vary dramatically depending on the spectral range, the optical properties of the cameras, the spectral characteristics of the scene, and the spatial resolution of the interested targets in the scene. A general purpose IR image fusion technique that is suitable for real-time applications is proposed. The proposed technique can support different scenarios by applying a multiscale detail detection and can be applied to images captured from different spectral regions of the spectrum by adaptively adjusting the contrast direction through cross checking between the source images. The feasibility of the proposed algorithm is demonstrated on registered multi-spectral and multi-focus IR images. Fusion results are presented and the performance of the proposed technique is compared with the baseline fusion methods through objective and subjective tests. The technique outperforms baseline methods in the subjective tests and provide promising results in objective quality metrics with an acceptable computational load. Besides, the proposed technique preserves object details and prevents undesired artifacts better than the baseline techniques in the image fusion scenario that contains four source images.

Autofocus of infrared cameras based on the cumulative probability of blur detection

The infrared (IR) cameras plays an important role in the measurement and analysis of object signature. However, especially the scientific IR cameras that are used for research and military purposes have manual focusing system that reduces the sensitivity and reliability of the measurement taken. Camera autofocus algorithms extract various features from the camera images in order to define a measure for determining the most focused camera image instance. In this work, a no-reference image quality measure is modified and the modified measure is proposed for the autofocus of infrared cameras. Experimental results show that the proposed measure can be used in the problem of autofocus of infrared cameras, successfully.

Compressive sensing applications for single detector seekers

In this work, the application of compressive sensing theory to single detector seekers is investigated. Compressive sensing is a novel signal processing technique which shows that a compressible signal can be constructed using fewer measurements obtained in a specific way below the Nyquist rate. Single detector image reconstruction applications using compressive sensing have been shown to be successful. Single detector infrared seekers suffer from low performance compared to infrared seekers utilizing costly focal plane array detectors. The single detector, pseudo-imaging rosette scanning seekers scan the scene with a specific pattern and process the resultant signal with signal processing methods to estimate the target location without forming an image. In this context, this type of old generation seekers can be converted to imaging systems by utilizing the samples obtained by the scanning pattern in conjunction with the compressive sensing theory framework. Images have been reconstructed from samples obtained by the rosette scanning pattern for different sample numbers and it has been shown that the results obtained are comparable to other sampling methods proposed in the literature.

Database based fourier coefficient sampling for sparse recovery of infrared sea surface platform images

Compressive sensing framework states that a signal which has sparse representation in a known basis may be reconstructed from samples obtained from a sub-Nyquist sampling rate. Due to its inherent properties, the Fourier domain is widely used in compressive sensing applications. Sparse signal recovery applications making use of a small number of Fourier transform coefficients, have made solutions to large scale data recovery problems, i.e. images, applicable and more practical. The sparse reconstruction of two dimensional images is performed by making use of sampling patterns generated by taking into consideration the general frequency characteristics of any image. In this work, instead of forming a general sampling pattern for infrared images of sea surface platforms, a special sampling pattern has been obtained by making use of a database containing images recorded under similar atmospheric conditions. It has been shown by experimental results that the proposed sampling pattern provides better sparse recovery compared to the widely used pattern proposed in the literature.