Target Detection Under The Blur Cloak Thing Based On The Known Knowledge

The target detection technology is extensive used in the realm of aerial-objects identification, ground-objects identification and detection of industrial product. Template matching technology is the key of target detection system. Matching namely that sensor captures two breadths of or many breadths of images aiming at in space versus identity scenery in different time and different image-forming condition, or according to known mode look for homologous mode in another breadth of image. The robustness of matching algorithm degrade owing to the reflects of some natural conditions, random texture, cloak thing like illumination and so on. The primary aim of this paper is that study the objects'detection possess random vein and lie to below the cloak-object. Selecting ceramic tile image which possess random vein and presence cloak-objects,missile target image and jungle moving target image as study objects.Mostly make research on matching algorithm 'robustness and under this three strain circumstance matching algorithm' accuracy and it's applicability. The work was done as follows:1. This paper analyses the key element of matching algorithm namely matching primitive, measurement function, search strategy, evaluation mechanism and in common use matching algorithm, and presents their applicable instances and their relative merits.2. Analyzing image feature of this paper, it possess random vein and the objects in a state of cloak, and studying matching algorithm. Due to the image exists the instances of spin and distortion, in order adopting part small template proceed matching in order to reduce thereof impacts by as possible, besides small template matching may decrease calculation quantity, and it is propitious to real time of matching. Looking into be on the impact of illumination, chromatism, under cloak circumstance template size and quantity select versus matching rate. Matching primitive chooses gray, gray information measure large and most easy to obtain,but it is rather effected by illumination condition and chromatism, Maximum matching rate restricted to 70%. Therefore, putting forward to amend matching primitive as gradient value that is lesser impacted by illumination and chromatism, and it's matching rate approve to 98%, but there is 1 loaf pixel error of part region. Measurement function selects least absoluteness difference and ML distance. The major factor of affected ML distance algorithmic is threshold value, and do further research for the selection of threshold value.3. There are three points through analyzing major cause this paper error brings, namely extreme point non-uniqueness, pickup camera head radial distortion and...