Research On Infrared Small Target Detection Algorithm Based On Image Denoising Technology

Infrared small target detection has been widely used in many areas such as infrared target tracking,precision guidance,remote early warning and so on.However,due to the infrared small target have the characteristics of small size,weak signal,complex background,etc.In the infrared imaging guidance system,the small target is disturbed by not only the background and the buildings of the image itself,but also the other serious noise interference,which can't be avoid in the process of acquisition or transmissionn,all above have brought difficulty to the target detection and tracking.Consequently,before target detection and tracking,it is necessary to preprocess each frame image obtaining in the infrared imaging guiding system to remove the interference of its background,buildings,clouds,noise and other disturbances.Infrared image de-noising is an important research content in digital image processing.Whether it can effectively remove the noise in the infrared image will directly affect the effectiveness of small target detection tracking,so taking the appropriate method for de-noising is a very important pre-processing step.There are many experts and scholars at home and abroad considering these difficulties,and they have put forward some innovative,robust and guided infrared detection and tracking of small target algorithms.Based on the previous research results,I have studied the detection algorithms of some small target tracking,and proposed two kinds of detection algorithms of small target.The main works are as follows:(1)In this thesis,a detection algorithm of infrared small target based on edge detection image de-noising is proposed,which aims to overcome the shortcomings of wavelet image de-noising method.Therefore,we fully consider the distribution characteristics and noise characteristics of Gaussian white noise.It is a de-noising method which combined with frequency domain and airspace.The performance of the algorithm is mainly determined by the accuracy of the threshold noise standard deviation estimation of wavelet coefficients and the performance of the noise function.Firstly,the algorithm estimates the noise variance of the image by edge detection.Then,the polynomial threshold function is used to de-noise.The selection of the polynomial threshold function is extremely critical to the noise attenuation and the protection of details in the mean times.Finally,the simple global threshold is used to cut up the wavelet reconstruction image of infrared small target detection which has been de-noised.The experimental results show that when de-noising,the proposed algorithm can protect the small target of the infrared image,which makes it possible to detect the infrared small target efficiently and accurately.It is a kind of infrared small target detection algorithm based on image de-noising technology with excellent performance.(2)This thesis presents an adaptive multimode infrared small target detection method based on wavelet transform.The algorithm can adjust the threshold according to the distribution of noise in different scales and directions,and effectively improve the accuracy and robustness of infrared small target detection in complex background.Firstly,the infrared image is transformed by two-layer discrete wavelet transform.It can 'be seen that the wavelet coefficients of Gaussian noise in the first-layer three directions are smaller,the second layer contains more infrared small target information and less noise,but the wavelet coefficients of the noise are large.Secondly,according to the distribution of noise,the first layer uses the space adaptive Bayes Shrink threshold to remove noise,the second layer uses the generalized cross validation(GCV)threshold for secondary de-noising,and the wavelet coefficients of the approximate sub-band are set to zero,then to start wavelet reconstruction.Finally,a simple global threshold is used to segment the image to complete the detection of infrared small target.