Research On Image Processing System Of Underwater Motion Platform Based On DSP

The underwater motion platform is a kind of unmanned underwater vehicle, whichplays a significant role in both military and civilian use. With the functions of theunderwater information acquisition, information analysis and information delivering, thelarge and medium-sized underwater vehicle has the following defects: large volume,complex function and expensive cost. Based on the microminiaturization of underwatervehicle and the requirement of image collection of underwater vehicle, the software andhardware system of underwater vehicle is designed under the core of digital signalprocessor. Besides, the thesis makes a research on the image processing algorithms in therealization of system software.The author probes into the research form three aspects: the systemic design of theunderwater motion platform, the realization of the underwater motion platform softwareand enhancement algorithm of underwater image and systemic image recovery of high saltand pepper noise pollution.At first, the thesis gives an introduction to the digital signal processorTMS320DM642and TMS320F28335after elaborating the development and applicationof underwater vehicle in military and civilian use at home and abroad. Moreover, theauthor illustrates the signal integrity problems in designing CCS and PCB used to programthe digital signal processor. The realization of the underwater motion platform isaccomplished through the exploration of DSP/BIOS program. Thus, the software willfunction from five basic ways. With the multi-threaded design, the whole program willprocesses from several independent tasks to make a quicker response and a morereasonable distribution of hardware resources.Then, the image enhancement algorithm is the key for the underwater motionplatform to effectively operating. Based on the features of the light traveling underwaterand the laws of imaging underwater, the homomorphic filter can effectively solve theuneven illumination and enhance the dark zone of the image without damaging the clearzone of the image, which is fundamental to further promote the image quality. Experiments are made to find that underwater degraded images with low contrast and poorinformation distinguishing can balance the image brightness, improve the blooming ofregional edge and effectively reinforce the specific information by using the orientationfilter algorithm.At last, the design has accomplished the recovery algorithm of systemic collecteddigital image, polluted by high spiced salt noise. Based on the thought of the directionalmedial weights, this algorithm recovers the image by the way of iteration. The contrastresult of the subjective and PSNR indicates that the algorithm can effectively recoverimages polluted by different degrees of high spiced salt noise. Moreover, the algorithmcan still achieve a good recovery even under the90percent spiced salt noise pollution.