Research On Dim Point Infrared Target Detection Under Complex Background

In recent years, along with the computer science and image processing timely fastdevelopment, the infrared image processing technology has been widely used innavigation tracking, artificial intelligence and automatic control in the field. Because theinfrared dim-small targets have far distance, the area is small, the form feature weak,signal-to-noise ratio low characteristic, the infrared image preprocessing technology andinfrared target detection techniques are difficult and important. How to improve theinfrared image preprocessing effect and moving targets detection accuracy becomes thekeys of infrared dim-small targets detection techniques, In infrared guidance, it hasgreat significance.First, this paper analyzes the causes of noise and the characteristics of the infraredimage under complex background, according to the complex background infrared imagenoise characteristic, we combine median filter and wavelet transform algorithm to filterthe mixture noise of infrared image. The algorithm firstly uses the median filter andthen the wavelet transform. The wavelet coefficient use median filter principle toproduce new coefficient matrix.Finally use the wavelet threshold denoising to generatethe new infrared image. The simulation results show that the new method can filter outthe complex background infrared image noise effectively, the new method is better thanthe traditional filter method. After the infrared image filtering process, we use the poorimage method and traditional light flow field algorithm to detect small dim targets,because the two existing algorithm are poor in real time detecting and the changingenvironment, we put the adaptive threshold value method in light flow field algorithm,even when the outside environment changes or infrared image target is weak and small.According to image’s grey value distribution,the adaptive threshold value methodwill divide the grey value of the infrared image into different levels, and eventuallycalculate the image of the optimal threshold value to achieve the dim small targets ininfrared image under the complex background.Finally, this paper use the Raymond SEED-VPM642to simulate the system. Thesimulation experiment results show that combining the median filter and wavelettransform method in the image preprocessing and basing on adaptive threshold value method of light flow algorithm can detect out dim small targets under the complexbackground, the algorithm is good at real-time performance and the changing ofexternal environment.At last,it achieves the ideal effect.