Dissertation
Dissertation > Industrial Technology > Radio electronics, telecommunications technology > Communicate > Communication theory > Signal processing > Image signal processing

Study on Morph Detection Theory and Technology for Dim Small Moving Target in Sequence Images

Author ChengDeJie
Tutor LiZaiMing
School University of Electronic Science and Technology
Course Communication and Information System
Keywords Dim Small Target Detection Morph detection Morph Segmentation Intensity Morph Motion Morph Mathematical Morphology Top-hat Filter Watedshed Transfomer Motion Parameter Association Estimation Track Association Estimation
CLC TN911.73
Type PhD thesis
Year 2006
Downloads 757
Quotes 8
Download Dissertation

Detection and tracking for dim small moving target in sequence images with complex clutter background are of vital importance to modern high-tech warfare.For long distance, the dim small moving target image is usually composed of several pixels without specific shape and texture.In this dissertation,are studied the intensity and motion features of the dim small moving target in sequence images, and is suggested its morph concept: the intensity morph for the target’s intensity feature, and the motion morph for motion feature. Then a new morph model for the detection of the dim small moving target is constructed.From the study on intensity morph of the target, the dim small target in scene image presents as some special gray hubble, and the detection of the special hubble is the key task for target detection in single image.The study on motion morph of the target shows that, the motion of the moving target in sequence images has their independent motion and the track of the target is continuous. The detection of the independent motion and track is key task for motion detection of the target.By analyzing the intensity and motion morph of the dim small target, the dissertation presents that the target detection is made only by intensity morph detection and motion morph detection together. Based on that, the morph detection theory of the dim small moving target is founded. Combining the morph detection theory with the traditional target detection methods, the morph detection methods for dim small target are developed, and two different morph detection schemes are suggested.A mathematical morphological top-hat filter is used in intensity morph detection to remove the background clutter and enhance the intensity hubble of the target to the noise in filtered image. The suspicious objects sets (SOs) are extracted by the threshold processing and hit or miss transformer, which is called the morph segmentation. More accurate segmentation of the SOs is obtained with the compulsive marking to the watershed segment by morph segmentation.After obtaining the SOs, some target-like interfence are removed by the motion parameter association estimation and motion tracking association estimation, and the Kalman filter is used to track the SOs, the real target is captured in the end.Several valuable conclusions achieved by the dissertation are listed as following.1. A new morph concept of the dim small moving target was suggested based on the study on intensity distribution feature and motion feature of the target in the sequence images, which including the intensity morph for the target’s intensity distribution feature, and the motion morph for the target’s motion feature. A new morph model for the dim small moving target detection was constructed.2. Based on the concept of morph for the dim small moving target, the theory of morph detection was founded, and the principle of the morph detection was suggested.3. Two morph detection schemes for detection of the dim small target were suggested: the scheme based on intensity morph and the scheme based on motion morph.4. Based on compulsive marking to the watershed segment with the morph segmentation, a new accurate segmentation method for the target was suggested.5. Two methods for false alarm suppression were suggested: the motion parameter association estimation method based onχ~2 test, and the motion track association estimation method.6. A robust weighted estimation method for noise suppression was suggested based on constant lost alarm ratio.7. The performance analysis of multi-frame spatio-integrated point target detection is presented. The condition of SNR and integration for target detection is studied.

Related Dissertations
More Dissertations