Statistical analysis of computer-generated thermal images based on overall modeling of line-scanning process

Proceedings of the 2004 International Conference on Quantitative InfraRed Thermography

Marincic²,

Więcek³

2004

Self Cite

IR remote sensing and reconnaissance applications require transmission and/or store of line-scan thermal (IRLS) images in real-time. IRLS images exhibit a high degree of spatial redundancy that a codec reduces optimizing trade-off between bitrate and image distortion. In this paper, we propose adaptive coding based on joint optimization of image segmentation scheme and operational codec parameters. In computer implementation, we achieved significant coding gain using a new image segmentation scheme and advanced JPEG2000 codec.

Section: Introductionmentioning

confidence: 99%

Terrain-adaptive infrared line-scan coding: a new image segmentation scheme

Proceedings of the 2004 International Conference on Quantitative InfraRed Thermography

Marincic²,

Więcek³

2004

Self Cite

“…Furthemore, the scanning of a thermal image involves a significant scanning area overlapping at great scanning angles, which does not exist in a standard video signal [1]. Our systematic studies of techniques for infrared image statistical data analysis, compression and efficient transmission, begun with construction of a new analytical/program model [2,3]. The obtained results were used as the guidelines for statistical analyses of real test images [4].…”

Section: Introductionmentioning

confidence: 99%

Efficient transform coding of IR line-scan images based on spatial adaptivity

Proceedings of the 2000 International Conference on Quantitative InfraRed Thermography

Marincic²,

Więcek³

et al. 2000

Self Cite

Based on our systematic studies of IRLS (InfraRed Line-Scanner) image statistical properties and comparative analyzes of state-of-the-art transform based image coders, we propose coding performance improvements by spatial image segmentation and operational rate-distortion optimization. We present in this paper, a new approach for an adaptive coding of IRLS images which combines the advantages of the following items: hierarchical tree-like image segmentation along-scan direction and Lagrange optimization of a trade-off rate-distortion D(R). The proposed discrete optimization algorithm is based on fast systematic search for the best possible image segmentation scheme, together with the best possible bit allocation in coding of each segment independently. We implemented proposed optimization algorithm on computer and obtained significant signal-to-noise (PSNR) improvement by spatial adaptivity and independent discrete wavelet transform (DWT) coding of each image segment as compared over the standard discrete cosine transform (DCT) coding of nonsegmented IRLS test images.

“…The terrain is sampled by a transverse scan as the aircraft moves forward along track. Consequently, IRLS operates as a linear angle scanning system, giving angular scan and linear displacement along track as the two dimensions of the still image [1]. The need for digital IRLS image compression arises from the fact that scanner generates enormous instantaneous bit rates which can easily saturate transmission link, the real-time display and human observer.…”

Section: Introductionmentioning

confidence: 99%

Acomparative study of advanced frequency-domain coding techniques in compression of infrared line-scan images

Proceedings of the 1998 International Conference on Quantitative InfraRed Thermography

Więcek²,

Marincic³

et al. 1998

Self Cite

A comparative study of transform coding techniques in data compression of IRLS (InfraRed Line-Scanner) images is described in this paper. In performance evaluation of block transform and wavelet image coders, we first carefully formed two sets of "good" basis functions: 4 types of generalized overlapped block transform (LOT) and 3 types of biorthonormal discrete wavelets transform (OWT). We found by extensive computer simulation on IRLS representative test-images, type and order of transform in the most efficient scalar quantized-entropy coder in each set for compression ratio from 10: 1 to 50: 1, based on objective and subjective quality measures of decoded test images. Finally, in performance comparison of the most efficient generalized block transform and wavelet coder we found that biorthogonal Ooubechies OWT based coder outperforms standard OCT and optimized LOT coders. Additionally, OWT coder gives possibility for further research on spatial image segmentation and terrainlaltitude adaptive IRLS image coding.