Research On Multi-resolution Analysis And Recognition Of Underwater Targets Acoustic Image

In recent years application regions of exploration of underwater acoustic imaging are more and more widely such as military target identification, seabed mapping, seabed oil exploration, bridge building, terminal building etc. Due to the complexity of the underwater sound field environment and the non-linear imaging of sonar equipment imaging, the collected underwater sonar images are of characteristics with low contrast, poor-quality imaging, low contrast between the target and the background, which have led to great difficulty to the follow work of the sonar target detection, identification and analysis. Multi-resolution analysis tool can multi-scale details analysis by basis function expansion, translation and other computing, and it can effectively extract information from signals. At present, Multi-resolution analysis is international acknowledged advanced technology in the domain of information and signal processing. Meanwhile it is the front a question for discussion and study hotspot. More and more people attach importance to the application of multi-resolution analysis in the domains such as signal filtering, image denoising, image fusion, image recognition etc.Multi-resolution analysis and its application in image processing are investigated mainly in this dissertation. The main work can be summarized on array signal denoising, the sonar image denoising, the sonar image fusion and the sonar image recognition.First of all, the working principle of underwater acoustic sensors of the front-end imaging equipment and acoustic array are general described and analyzed. Due to the complexity of background noise, the received signal is usually polluted, which will affect availability and reliability of goal detection and classification results. The denoising method of underwater multi-sensors array signals is proposed based on the Surfacelet transform. Tests on acoustic signal denoising show that the proposed method can realize the parallel calculations for array signals, and the denoising effect is satisfactory. The method takes use of not only relativity of underwater array signals, but also translation invariant of Surfacelet transform, which makes the detection and recognition of underwater goal signals be accurate. The method is an effective method.Secondly, the concepts and denoising methods in common use multi-resolution transform are recapitulative introduced and analyzed. At the same time, the imaging mechanism and the statistics characteristic of underwater acoustic image is expatiated:Through studying acoustic image of character which is bad contrast and deteriorates edges and detail, two the acoustic image denosing methods based on multi-resolution analysis are proposed. They are Sonar image denoising method based on NSCT circle sample and the method for image denoising based on the Surfacelet transform and multidirectional Cycle Spinning respectively. These methods are characterized by multi-direction option, parallel information processing, high efficiency of information using, and fusing the enhanced on multi-resolution. By the simulation results, the effectiveness and superiority of the methods are proved.Thirdly, in order to fuse digital information of MBES and SSS, the sonar image fusion method based on Surfacelet transform is researched. The method can extract image borderline and detail region in effect by using Surfacelet coefficient trait. The fusion course will classify coefficients, and then the different fusion method is used to high frequency and low frequency respectively, which takes full used of value information. The low-frequency sub-band uses a linear average fusion; the band-pass is divided into details areas and non-detail areas, which are used different fusion rule; the high-frequency part fuses to reference window energy value. The method not only improves information quantity and definition after fusion underwater acoustic image, but also makes the best of translation invariant of Surfacelet transform and cannot output result local distortion during the course of fusion. The fusion processing based on Retinex model to the decomposed acoustic image can availability use of dark region information of the underwater acoustic image. The proposed method apply to MBES and SSS acoustic images, which has obviously better effect than maximum, Laplacian, contract fusion and method on basis of Wavelet transform.Finally, in common use the image character extraction technology and pattern recognition methods are recapitulative introduced and analyzed. Aiming at seabed texture information traits provided high resolution underwater acoustic image, take appropriate algorithms to underwater acoustic image recogniton based on multi-resolution tools with texture as substrate features. The acoustic image texture recognition method is designed based on the Retinex model and Surfacelet transform, which makes full use of dark areas information of acoustic image. The experimental results show that the recognition rate of the method is better than others.In general, this dissertation had researched on multi-resolution analysis and its application on the underwater acoustic image processing based on multi-resolution analysis. Aiming at the shortcoming of the region, the improved algorithms were proposed. Experimental results show that the adopted intelligent optimization algorithms and the proposed projects could attain good results.