Practical transform coding of multispectral imagery

We present an implementable three dimensional terrain adaptive transform based bandwidth compression technique for multispectral imagery. The algorithm exploits the inherent spectral and spatial correlations in the data. The compression technique is based on Karhunen-Loeve transformation for spectral decorrelation followed by the standard JPEG algorithm for coding the resulting spectrally decorrelated eigen images. The algorithm is conveniently parameterized to accommodate reconstructed image fidelities ranging from near-lossless at about 5:1 CR to visually lossy beginning at about 30:1 CR. The novelty of this technique lies in its unique capability to adaptively vary the characteristics of the spectral correlation transformation as a function of the variation of the local terrain. The spectral and spatial modularity of the algorithm architecture allows the JPEG to be replaced by a alternate spatial coding procedure. The significant practical advantage of this proposed approach is that it is based on the standard and highly developed JPEG compression technology     

Adaptive Transform Coding Based on Chain Coding Concepts

Chain coding technique, originally developed for digital representation and processing of line drawing data, has been implemented in a transform image coding algorithm with significant performance improvement. The algorithm is based on the observation that the boundary of the regions of zero coefficients within a transform block can be efficiently represented by sequences of fixed line segments (chains). Preliminary results indicate significant improvements over the basic coder algorithm in which the consecutive zeros in the transform block were runlength coded. The additional implementation complexity is modest.     

Detection of moving point objects against a moving background

The temporal spectral characteristics of a dim moving point object and a moving background, as observed by a sensor array, are analyzed. This type of problem occurs in remote sensing, machine vision, and many other applications. The diffraction limitation of the sensor optics ensures that the temporal spectrum of the background moving with a finite velocity has a finite maximum bandwidth, regardless of background structure. Because the outputs of the sensor array are time sampled, its spectrum is infinitely replicated over an interval of temporal frequency equal to the reciprocal of the sampling time. If this interval is at least twice as large as the maximum background temporal frequency, there is a region with no background components in the middle of each interval. However, because the point object temporal spectrum in the sampled sensor array output is continuously distributed, this region will contain part of the point object signal. Thus, a criterion for the existence of an effective background suppression filter is that the point object fundamental frequency must be greater than the maximum background temporal frequency. When this criterion is satisfied, the amount of background leakage in the filter depends on the sharpness of its passband response and its stopband characteristics. In general, higher-order filters have sharper response and hence better performance. If the criterion is not met, all types of filter lose their effectiveness since the background signal will leak through the passband of the filter. The fundamental concepts developed here were examined for some typical parameter values. It is shown that for this system the point object can be effectively discriminated. In some cases the point object and background temporal spectral responses vary significantly with spatial position within the field of view. Because the filter's center frequency must match the point object temporal fundamental frequency, it is necessary to use an adaptive filter in these situations.     

Analysis of the precision of generalized chain codes for the representation of planar curves

This paper examines a set of line-segment approximation codes for the representation of planar curves (the so-called generalized chain codes) and shows that the average quantization error (measure of code's precision) is directly proportional to the grid size and is independent of the form of the code. Thus, to achieve a desired level of precision for the representation of a line drawing, only the size of the grid need be determined; the form of the code can be chosen on the basis of other criteria, such as compactness, smoothness, or relative ease of encoding and processing.