This paper proposes to use empirical mode decomposition (EMD) with spectral gradient enhancement increase the classification accuracy of hyperspectral images support vector machine (SVM) classification. Recently, it has been shown that higher image can be achieved by using 2-D EMD is applied each band separately obtain intrinsic functions (IMFs) band, while sum IMFs are used as feature data in SVM process. In previous approach, have summed directly, i.e., equal weights. It this paper,...

Hyperspectral imaging provides high spectral resolution and thereby improved classification, detection, recognition capabilities with respect to standard systems. However, hyperspectral images generally have low spatial resolution, varying from a few tens of meters, resulting technical limitations such as platform data storing capacity satellite-to-ground transmission bandwidth. Spectral unmixing information on pixels in terms abundances pure signatures, without providing distribution at...

Hyperspectral images provide significant information about the chemical constituents of materials and increased classification recognition capability, due to high content spectral they contain. Therefore, hyperspectral imaging has found use in many application areas, including recycling. The spectra paper, cardboards, glass, metal plastics can be measured used for their identification, consequently sorting, different properties. In this study, various recycling material samples are...

This paper targets hard-cut detection for archive film, i.e., mainly black-and-white videos from the beginning of last century, which is a particularly difficult task due to heavy visual degradations encountered in sequences. A robust system based on modified phase correlation presented. Phase-correlation-based carried out using spatially sub-sampled video frames, and candidate indicated case low correlation. double thresholding approach consisting global threshold used conjunction with an...

This letter proposes the use of denoising in conjunction with 2-D empirical mode decomposition (2D-EMD) hyperspectral image bands for higher classification accuracy. Initially, 2D-EMD is performed to into intrinsic functions (IMFs). Then, applied first IMF each band because this includes local high-spatial-frequency components. Features reconstructed as sums lower order IMFs are then used classification. Support vector machine a approach letter. Experimental results show that proposed...

In this paper a novel adaptive fuzzy: image sequence stabilization system is proposed for the motion correction part of digital stabilizers. The stabilizer based on smoothing absolute frame positions. Initially short mean filter applied to raw displacements as pre-process, reduce dynamic range fuzzy input. This enables give appropriate responses with fewer number membership functions, hence provides improved performance at reduced computational load. Fuzzy then achieved through mapping....

This paper presents an automatic scratch detection and removal approach for archive film sequences. The proposed detector mainly exploits temporal coherency of candidate positions, which are obtained using method in the literature. In restoration stage, both spatial information employed. edge priority based algorithm successfully removes effects from

Block based motion estimation (ME) techniques on low bit-depth representation have been proposed in the literature to reduce computational complexity of ME. One-bit transform ME methods are pioneers this field. These derive one-bit images using an adaptive thresholding scheme, and perform block simplified matching criteria. Recently, constrained (C-1BT) has increase performance 1BT ME, restraining pixels with values close threshold throughout matching. In work, a weighted constraint strategy...

Image sequence stabilisation based on fuzzy adaptive Kalman filtering of absolute frame displacements is proposed. Two filters are operated in parallel, one which used as a reference filter and run with constant high process noise variance to enable smooth close tracking global camera displacements. The the adaptively varied according residual between output.

Hyperspectral image compression is an important task, where the aim to store or transmit data in efficient way. images are mostly captured by a sensor system that includes multiple imaging sensors covering different regions of electromagnetic spectrum. Misalignment produces boresight effect, and this problem can degrade band prediction performance noticeably. Another blurry images. In order gain robustness these problems, bimodal conventional recursive least-squares (B-CRLS) method proposed...

In this paper, a system that converts Turkish signs to words using convolutional neural networks is presented. Skeleton data obtained by Microsoft's Kinect device used in the proposed system. User located front of sensor makes corresponding word for limited time. Afterwards, skeleton joints are extracted. Finally, points on consecutive frames merged, and Word classification performed convulational network. experimental studies, evaluations carried out with our own dataset performance method...

In this paper a new mine detection algorithm for side scan sonar images is proposed. the proposed method, an illumination normalization to compensate variations based on Discrete Cosines Transform applied initially images. Then, segmentation carried out using Markov Random Fields and potential mines are detected.

Hyperspectral imaging provides increased capability for many image processing tasks with respect to standard systems. One of such in hyperspectral is change detection, which aims detect the differences occurring between images acquired from same scene at different times. In this paper, a panoramic system used capture multitemporal data, and novel multi-band Census Transform (MCT) proposed detection on these data. Experimental results validate performance method utilized acquisition system.

In this letter, a modification to phase–correlation‐(PC‐)based supervised classification method for hyperspectral data is proposed. An adaptive approach using different numbers of multiple class representatives (CRs) extracted PC‐based k‐means clustering each compared with the use selecting small, pre‐determined number dissimilar CRs. PC used as distance measure in determine spectral similarity between pixel and cluster centre. The chosen adaptively, depending on training samples class....

The inherent chemical properties of materials can be brought into perspective using the large amount spectral information provided by hyperspectral imaging systems. Therefore, utilization in industrial applications is gradually increasing. One sectors that benefit from advantages recycling. Plastics which have different (PP, PE, PVC, PET and PS) need sorting for plastic waste In this study, types plastics images acquired a shortwave infrared (SWIR) system are successfully sorted.

Images can have poor visibility, contrast and colors in foggy weather conditions. Therefore it is required to enhance visual quality of the fog-degraded images. In this paper we present a new method based on an Empirical Mode Decomposition (EMD) for image enhancement. Initially each spectral component decomposed into Intrinsic Functions (IMFs) using EMD. Then enhanced constructed by combining IMFs channels with optimum weights order obtain increased quality. The optimal weight estimation...

Spectral unmixing is the process of identifying pure spectral signatures, called endmembers, from a hyperspectral data, and then expressing each pixel vector in terms fractional abundances these endmembers. Most endmember extraction methods literature use only information, whereas spatial composition data disregarded. Spatial preprocessing methods, that are motivated by assumption endmembers more likely to be located homogeneous regions instead transition areas, can alleviate this drawback...