Amir L. Liaghati

An efficient method for lossless compression of bi-level ROI maps of hyperspectral images

While one can achieve very large size reduction on a hyperspectral image dataset by preserving only some regions-of-interest (ROIs), the bi-level map that describes the locations of the ROI pixels tend to defy efficient compression due to the somewhat “random” nature of ROI pixel locations. To this end, we proposed a novel method for lossless compression of these ROI maps. In this method, we first partitioned a bi-level map into equally sized blocks. We then converted the bi-level pixels within each block into a block symbol. Based on the observation that the most probable blocks tend to contain either all zeros or all ones, we chose to run-length code these most probable block symbols before applying Huffman code in order to achieve high compression, whereas we applied a separate Huffman code on other less probable block symbols. Thus this biased run-length coding method differs from conventional approaches where all symbols are run-length coded. Tests on NASAs AVIRIS dataset showed that the proposed method could provide significant improvements over various bi-level image compression techniques (including JBIG2 and lossless JPEG 2000) on the ROI maps.

Improved distance coding of binary images by run length coding of the most probable interval

We proposed a new method to improve our previous work on efficient distance-coding of binary images, where we compressed a binary image by applying the bzip2 lossless data compressor on a sequence of intervals, which represent the distances between identical source symbols - either zeros or ones for binary images. Motivated by the observation that a majority of intervals tends to be one, we propose to run-length code this most probable interval independently from the rest of the intervals. Separate Huffman coding tables were used to code the run-lengths of the most probable interval versus other intervals. Consequently, this hybrid coding scheme allows a fraction of one bit to be assigned to the most probable interval on average, as opposed to at least one bit per interval without run-length coding, thereby contributing to about 17% improvement on the compression ratios on some test images.

An adaptive DPCM method for efficient data compression in aerospace sensor systems

In this paper, we explore different techniques for lossless compression of sensor data on a wireless sensor network. In this work, we applied modify Differential Pulse Code Modulation (DPCM) scheme for our baseline lossless compression. These techniques were applied on sensor types with reasonably predictable data pattern (e.g., temperature sensor, pressure sensor, etc.) allowing about 50% data compression on the original samples. In addition, we propose a new adaptive data compression algorithm based on DPCM for higher compression of sensor data. Furthermore, this work proposes a new mathematical model based on Markov model to show where the proposed algorithm is beneficial over the conventional DPCM methods.

An adaptive interval generation method for efficient distance coding of binary images

We proposed an adaptive method for more efficient distance-coding of binary images. The proposed method partitions the image into blocks where the interval sequences of zeros or ones can be calculated, as opposed to the conventional method where intervals are calculated by following a fixed scan order. In the proposed method, one can adaptively choose either horizontal or vertical scan within a block, depending on criteria based on entropy values. The resulting intervals tend to have lower entropies than the conventional non-adaptive methods, thereby allowing for higher compression than distance coding using a lossless codec. Our simulations on various test images demonstrated that (i) the proposed method achieved significantly higher compression than the non-adaptive distance coding method; (ii) the proposed method can be used as an efficient preprocessor of a lossless coder, offering higher compression than directly coding on the original images.

A probabilistic time variant sensor accuracy model and GUI in aerospace applications

In this work, we proposed a unique model and Graphic User Interface (GUI) to measure the sensor accuracy with different accuracy regions for aerospace applications. Sometimes the measurement region for the sensor is a wide range, and the sensor might have higher accuracy in different regions. In other words, the accuracy values are not uniform for the entire range. In this work we developed a Matlab model and GUI that estimate the accuracy of the overall performance of the sensor when there are multiple accuracy regions of differing tolerances. This model and GUI use probability distribution parameters as a feedback input and the error distribution of the sensor to estimate the accuracy during each flight/ground phase/time interval which makes it a time variant model. The proposed GUI can be used to display multiple sensor error functions for different user inputted probabilistic parameters from estimated measurement distributions. This model and GUI can be used to determine what sensors to be selected, where the sensors can be used, and guide a machine or a user for a more productive decision making process. The current model and GUI exist in Matlab and return error curves for different probabilistic parameters to display the effect of changing your parameters due to expected distribution of measurements. It introduces a high fidelity model that could meet challenging program accuracy requirements, reduce the cost of man power, and processing time. Moreover, the proposed tool can be used to estimate the error generated by the resistance of the cable in a two-wire Resistance Temperature Detectors (RTDs) and predicts the behavior of error for different lengths of cable. This can help us use two-wire RTDs instead of three-wired RTDs where the probability distribution of measurements is outside the sensitive region in the extreme cold (cryogenic) region due to the delta resistance between temperature measurements being very small. This reduces the mass which significantly reduces the cost and installation time as well.

Evaluation of the biased run-length coding method on binary images generated by a modified Ising model

In this work, we evaluate the performance of several lossless compression methods on binary images simulated by a statistical model. This work was motivated by our previous study on compression of regions-of-interest maps in hyperspectral images, where a biased run-length coding method was found to provide good compression on just a few binary images with most pixels being in the background. In order to discover the trend on the compression performance, we introduced a modified Ising model of Markov random fields by Metropolis, which can generate a large number of binary images iteratively, so that we can study how the compression ratios of the above method change gradually when the contents of the images get varied slightly with iterations. Simulation results showed that the biased run-length coding method significantly outperformed the arithmetic code and JBIG2 method. The study also gave rise to a dual biased run-length method, which can provide further compression gains on images with more foreground objects.

A novel scheme for telemetry system data rate optimization

Limited telemetry bandwidth due to restricted radio frequency spectrum allocation is typically one of the most challenging problems when designing a telemetry system for space applications. To further complicate the problem, a large percentage of the allotted bandwidth is consumed by the over-head required by each packet to follow various standards and layer of protocols used in the telemetry system. This results in inefficiency in the actual telemetry data downlinked to the ground station. In the typical telemetry design, only one virtual channel is used per packet. As a result, in order to achieve the required packet size, filled data is inserted into the remaining bits. This Idle Packet consists of a set pattern of binary digits, and is considered part of the overhead, as its sole purpose is to fill the packet. In this work, a new scheme is proposed which takes advantage of the empty bits by starting the next virtual channel instead of inserting filled bits. This method will reduce the overall amount of overheard, in addition to allowing more data to be transmitted.