Research On The Application Of Sonar Imaging Based On Compressed Sensing

Compressed Sensing is a new recent method which is different from Nyquistsampling theorem essentially, and attract many researchers’ attention, especially inthe application of sonar imaging, super resolution image reconstruction, targetdetection and so on. The bottleneck of imaging systems refers to the sampling cost,however, Compressed Sensing offers the promise of reducing the amounts ofsamples, since only compressed data are gathered. As a result, the study plays animportant role in modern world.Firstly, we discuss the Compressed Sensing Theory and its application in Sonarimaging, introduce the basic framework and mathematical model of CompressedSensing, analyze the advantages of imaging technology, and Compressed Sensing isbrought in Sonar imaging as a necessity. In the simulation, back projection andCompressed Sensing are adopted in the virtual scene of sonar imaging, andcompared in four aspects of measurements, reconstruction time, image quality andcomplexity.Then, Compressed Sensing theory is applied in the area of super resolutionimage reconstruction since there is a desired problem of low resolution. Similarly,there exists a brief presentation of essential model. Compressed Sensing method iscompared with bicubic interpolation, neighbor embedding, iterative back projectionand so on, which employs the assessments of Structural Similarity Index, StructureContent and Spatial Frequency, and the influence of CS parameters is discussednext.Finally, background and target dictionaries are formed by background andtarget training samples separately, and the objects of the test images are identifiedby these two dictionaries, whose results are compared with other two algorithms.