Saliency-guided compressive sensing approach to efficient laser range measurement

TitleSaliency-guided compressive sensing approach to efficient laser range measurement
Publication TypeJournal Article
Year of Publication2013
AuthorsSchwartz, S., A. Wong, and D. A. Clausi
JournalJournal of Visual Communication and Image Representation
Date Published02/2013
Keywords3-D Data Reconstruction, Compressed Sampling, Compressive Sensing, Laser Measurements, Range Data Acquisition, Range Measurement, Saliency, Sparse Measurements Model

The acquisition of laser range measurements can be a time consuming process for situations where high spatial resolution is required. As such, optimizing the acquisition mechanism is of high importance for many range measurement applications. Acquiring such data through a dynamically small subset of measurement locations can address this problem. In such a case, the measured information can be regarded as incomplete, which necessitates the application of special reconstruction tools to recover the original data set. The reconstruction can be performed based on the concept of sparse signal representation. Recovering signals and images from their sub-Nyquist measurements forms the core idea of compressive sensing (CS). A new saliency-guided CS-based algorithm for improving the reconstruction of range image from sparse laser range measurements has been developed. This system samples the object of interest through an optimized probability density function derived based on saliency rather than a uniform random distribution. Particularly, we demonstrate a saliency-guided sampling method for simultaneously sensing and coding range image, which requires less than half the samples needed by conventional CS while maintaining the same reconstruction performance, or alternatively reconstruct range image using the same number of samples as conventional CS with a 16 dB improvement in signal-to-noise ratio. For example, to achieve a reconstruction SNR of 30 dB, the saliency-guided approach required 30% of the samples in comparison to the standard CS approach that required 90% of the samples in order to achieve similar performance.