A Method of Infrared Weak Small Target Detection Using Directional Derivatives to Construct Entropy Contrast

Abstract

The present invention relates to a detection method for small and weak infrared targets using directional derivatives to construct entropy contrast. It has four steps: first, based on least squares surface fitting, using a facet model to quickly solve each pixel point of an infrared image by using a convolution operation The first-order derivatives in four directions respectively; then, starting from the mathematical imaging model of weak and small targets, based on the directional derivative features and combined with the gray distribution information, the information entropy calculation formula is redesigned in the central area of ​​the small-scale neighborhood; then, For the surrounding area of ​​the small-scale neighborhood, the information entropy calculation formula is also redesigned, and the entropy contrast is constructed by taking advantage of the strong contrast between the central area of ​​the weak target and the surrounding area, and the background is suppressed in the derivative sub-map in each direction by using this measure. The target is enhanced; finally, the entropy contrast of the derivative sub-images in each direction is multiplied and fused to further suppress the edge clutter background and highlight weak and small targets. The invention can be widely used in the weak and small target detection of infrared images, and has broad market prospect and application value.

Description

[0001] (1) Technical field [0002] The invention relates to an infrared weak and small target detection method using directional derivatives to construct entropy contrast, belongs to the field of digital image processing, and mainly relates to facet models and weak and small target detection technologies. It has broad application prospects in various image-based application systems. [0003] (2) Background technology [0004] In the infrared-related early warning system, some suspicious targets such as ships, aircraft and missiles appear as bright areas on the infrared image due to their high temperature characteristics. Early detection of these suspicious objects can take countermeasures in advance to provide for subsequent strategic deployment, etc. Buy your precious time. However, when the imaging distance is long, these objects become some weak point targets in the image, and lose the features such as structure and texture that can be used. In addition, in contrast, the i...

AI Technical Summary

Problems solved by technology

Based on some small target detection methods based on morphological processing, such as hit-to-miss transformation, top-hat transformation, etc. : 43(7): 1084-1090. (Bai X, Zhou F. Hit-or-miss transform based infrared dim small target enhancement[J]. Optics&Laser Technology, 2011, 43(7): 1084-1090.)) Utilization Hit-to-miss transformation is used to enhance weak targets and suppress complex backgrounds. Mathematical operations based on morphology are relatively simple and fast, but they are easily interfered by noise and edge clutter and cause excessive false alarms.

A detection method based on the human visual system (see literature: Han Jinhui et al., A small infrared target detection algorithm based on the human visual system, Institute of Electrical and Electronics Engineers Geology and Remote Sensing Letters, 2016, 13(3):452-456.( Han J, Ma Y, Huang J, et al.An infrared small target detecting algorithm based on human visual system[J].IEEE Geoscience and Remote Sensing Letters,2016,13(3):452-456.)) using DoG to enhance the target Suppress the background and increase the direction information to overcome the disadvantage of DoG being sensitive to edges, and reduce the false alarm in the detection results. However, because some background edges in the infrared scene are not only along a fixed direction, the edge suppression effect is limited.

Considering that the target may have varying shapes and sizes when constructing local contrast measures, some methods add multi-scale information (see literature: Wei Yantao et al., Infrared Small Target Detection Based on Multi-scale Block Contrast Measure, Pattern Recognition, 2016,58 :216-226.(Wei Y,You X,LiH.Multiscale patch-based contrast measure for small infrared target detection[J].Pattern Recognition,2016,58:216-226.)), results from multiple scales Select the appropriate one as the final result, but the local contrast method itself is sensitive to noise, and the simple maxima-minimum scale selection method cannot avoid the influence of noise, so there are still many clutter false alarms in the final result

Most algorithms are still unable to obtain satisfactory results in target detection under low signal-to-noise ratio and false alarm suppression under complex background

Images