Research On Dim Small Targets Of Automatic Detection Technology Robustly And Timely Under Low SNR And Complex Background

Now, study on dim small targets of automatic detection technology robustly and timely under low SNR and complex background is still a challenge. The main difficulties are that a small target just occupies very few pixels in low signal-to-noise ratio infrared images, and image sequences have high frame rate and image resolution are high. Especially when the image sequences are high frame rate, and images have high resolution, the traditional algorithm is difficult to meet the system real-time performance, accuracy, stability and robustness requirements.In this article, we study the property of the Top-hat transform, and explore the method Bai. Et proposed to import a judging value t to different the real target and false target in the infrared image. Then, a novel method is proposed which is called modified white dilation algorithm to enhance the target in the infrared image. Firstly, a ring structure element is formed using the property of small target region. Secondly, a dilated image is obtained using the ring structure element. Finally, the dilated image and the original image is compared to obtain a difference image. Experimental results verified that the modified dilation algorithm for target enhancement under the conditions of heavy clutter and dim target intensity was efficient, effective and robust.Then, according to the concept of Shannon entropy, local histogram entropy map is proposed combining the concepts of local entropy with image entropy, which is to solve that problem by enhancing small targets. Then, this article analyzes the property of the local histogram entropy map, experiments show that the local histogram entropy has a strong edge detection. Finally, LHEM is used to detect small targets in images with different background, and both quantitative analysis and qualitative comparison confirm the validity and efficiency of the presented approach.