JPH07306939A - Exclusion method of clutter by making use of connectivity - Google Patents

Abstract

PURPOSE: To provide a method for isolating a small target object from a background clutter by using a connectivity algorithm. CONSTITUTION: In an FLIR target detection system, a true target 52 is isolated from a latent erroneous target indication generated by a background clutter by using a connectivity algorithm, and this algorithm tries to find a closed path 65 with a low or high intensity within a scheduled distance from a centroid 64 of the latent target around each target 52. When the trial is successful, the latent target is the true target, and otherwise, it is the clutter. The erroneous target indication can be extremely reduced in this method.

Description

Detailed Description of the Invention

[0001]

FIELD OF THE INVENTION This invention relates to a method for clutter elimination in digital images, and more particularly to a method for isolating small target objects from background clutter by using a connectivity algorithm.

[0002]

2. Description of the Prior Art In forward looking infrared (FLIR) imaging of non-uniform scenes such as terrain, point targets where the temperature is significantly higher or lower than the background temperature (for long spreads). Aircraft) is often not detected by the target detection algorithms conventionally used in FLIR imaging. Also, conventional target detection algorithms may record many background characteristic clutter as targets. Target detection algorithms of this type include leading edge-trailing edge detection and least mean square (LMS) filters. One-dimensional LMS filters can produce false targeting at the edges of background objects, as is the case with leading-trailing detection, and even using two orthogonal one-dimensional LMS filters, In addition, an incorrect target indication may occur at the corner. In order to isolate the point targets for a clear display, it is necessary to determine the spatial extent and spatial connectivity of the relevant objects. This is a difficult calculation technique.

Prior art in this field concerns Ma image US Pat. No. 4,742,557, which shows an apparatus for extracting symbolic information from a noisy background in an optical scanner, preprocessing an image by median subtraction. U.S. Patent No. 4,
Covers 736,439 and methods for analyzing edge spectra
Choate in U.S. Pat. No. 4,685,143.

[0004]

In accordance with the invention, an imaging electronic device tests the pixel intensity of pixels forming a closed path (65) around a centroid (64) of a potential target object (52). To do. The electronic device under test requires that its intensity differ from the intensity of the centroid in a predetermined manner in order for the pixel to form part of the closed path 65. If this type of closed path 65 can be drawn within a predetermined distance of any pixels of the closed path from the centroid, then the potential target goal 52.
Are identified as true targets and can be displayed as such.

In order to facilitate the necessary calculations in real time,
The present invention provides a test pattern (66) that minimizes the number of strength comparisons required to determine the presence of a closed path. This is done by performing a comparison on the rectangular path (20-26) and the radiation pattern. Thus, each radiation ends as soon as a sufficient intensity difference is found on the line.

[0006]

DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 shows a representative FLIR image 50. Most pixels in the display are prescreened before the connectivity clutter elimination of the present invention is performed. The target aircraft is shown at 52 in FIG.

FIG. 2 illustrates, in block diagram form, an exemplary FLIR target positioner (locator) using the present invention. The infrared image seen by the sensor 54 is digitally processed by the imaging circuit 56 and stored in the frame memory 58 one frame at a time.

Algorithms such as Least Mean Square (LMS) are conventionally used in the target locator 60 to prescreen pixels to find centroids of interest in the image 50 which have the attributes of potential targets. Has been done. The nature of these algorithms is such that objects can be identified as potential targets even if they are connected to large objects or they are too slender and not actual targets. As a result, processing the image 50 to the target display often results in false targets. This false target identification is
Known as clutter. Each potential target in the image 50 is then tested by the clutter elimination device 62 of the present invention to eliminate false targets and display only true targets (in an improved form) on the target display 63. In the device of FIG. 2, all elements except the clutter rejection device are conventional and well known.

In accordance with the present invention, the centroid of potential target 52 is identified in image 50 by conventional target-positioning algorithms. An attempt is then made to find a closed path, such as path 65 in FIG. 3, which surrounds the centroid 64 without passing through any pixels having the attributes of the potential target (hatched area in FIG. 3).

In a digitized image, such as image 50,
The centroid pixel 64 can be considered to be surrounded by a series of closed arc-shaped closed paths. These closed paths may or may not appear as in FIG. 1, but are shown in more detail in FIG. In this figure, the centroid pixel 64 is the pixel 11 × of the digitized image 50 stored in the memory 58.
Eleven (as an example) in the center of array 66. Array 66
Are pixel paths 68, 70, 7 as shown in FIG.
2, 74 and 76.

A potential target is considered to be a true target because a virtual closure line can be drawn around the centroid 64 within the distance that the virtual closure line was scheduled, and is therefore considered to be a true target, so that all in the path. Pixel intensity + positive threshold for the point
Is less than the intensity of (for the hot target) and the pixel intensity-positive threshold is greater than the intensity of the centroid 64 (for the cold target), then it is considered to be a true target. Threshold is a sensitivity factor used for clutter rejection.
In the embodiment of FIG. 4, the present invention provides routes 70, 72, 7
Detect the presence of such a path by testing 4,76 pixels in that order (path 68 is not tested because, assuming the target is at least 3 × 3 in size, this path is: Because it approaches centroid 64 as it would always be assumed to form part of the target).

In the concatenation algorithm used in the preferred embodiment of the present invention, the pixels in path 70 are first represented in FIG.
They are tested in the numerical order indicated in. If it is found that one of these pixels has an intensity greater than the intensity of the centroid 64 minus the positive threshold (for hot targets), no flag is set in memory. The pixels in path 72 are then tested in the numerical order shown.
However, if pixels with the same number were flagged in the previous path (ie path 70), it will be flagged without testing. If a path is found in which all pixels are flagged, then the potential target passes the test. Otherwise, the process is repeated on path 74 if none of the pixels in 72 are flagged, and finally on path 76 if none of the pixels of 74 are flagged. A flow chart illustrating the process is shown in FIG.

The paths 70, 72, 74, 7 shown in FIG.
The test order of the pixels in 6 is preferably such that pixels with the same order number in the sequential path are centroids 6
4 forming a radiation pattern radiating from 4 and the sequence numbers in each path are arranged in a sequential number pair arranged on opposite sides of the diameter of the centroid 64. This array is set by examining the target data and determining which array has the highest possible statistical probability of finding a closed path (if one exists) in the shortest possible time.

Upon completion of testing of path 76, or any of the earlier paths, if all pixels in the path are flagged, a lower strength closed path within the allotted distance from centroid 64 will be created. Targets that are present and centered on the centroid 64 are identified as true.
If any unflagged pixels remain in the final path 76, the target is identified as false.

Four paths were tested in FIG.
The number of paths can be more or less to accommodate a particular imaging technique or a particular type of target parameter,
Their distance from the centroid 64 can be changed. For this purpose, it is beneficial to visually display the route as shown in FIG. For the same reason, the paths may be separated from each other and may be other than rectangular without departing from the spirit of the invention.

Although the present invention has been described with reference to its preferred embodiments, it should be understood that the invention is susceptible to various changes and modifications without departing from the spirit of the invention.

[Brief description of drawings]

FIG. 1 is a FLIR image showing a target.

FIG. 2 is a block diagram of a device on which the present invention operates.

FIG. 3 is a diagram illustrating a closed path around a target object.

FIG. 4 is a diagram showing an example of nested pixel paths and their pixel comparison order.

FIG. 5 is a flowchart showing the functions of the present invention.

[Explanation of symbols]

 50: Forward monitoring infrared image 52: Target aircraft 54: Sensor 56: Imaging circuit 58: Frame memory 60: Target locator 62: Clutter elimination device 63: Target display 65: Path 66: Test pattern 68, 70, 72, 74, 76: Pixel path

Front Page Continuation (72) Inventor Robert Sea Gardemar Jr. 10 Sangallo, Irvine, CA, USA

Claims (8)

[Claims] 1. Finding a centroid pixel of a potential target of interest in an image, establishing at least one closed path of pixels around this centroid pixel, 1 intensity of individual pixels of said path to said centroid The method of clutter rejection in a digitized image comprising the steps of identifying the potential target object as true if all of the individual pixels have a predetermined intensity for the centroid pixel as compared to the intensity of the. 2. A potential target when a plurality of nested paths of pixels are established around the pixel and all pixels of any path have the predetermined intensity for the centroid pixel. The clutter rejection method of claim 1, wherein the target is identified as true. 3. The clutter elimination method according to claim 2, wherein the routes are compared with the centroid one by one at a time, and the process proceeds from the innermost end of the nested route to the external report. 4. The clutter exclusion of claim 3, wherein the pixels are compared in numerical order such that the same sequence number in successive paths forms a pixel line that radiates substantially outward from the centroid pixel. Method. 5. Finding a centroid pixel of a potential target object in the image, establishing a plurality of closed paths of pixels around the centroid pixel, and moving outward from the innermost end of the path. like,
The intensities of the individual pixels of the path are compared to the intensities of the centroid, and the comparison forms pixels lines that radiate outwardly from the centroid pixels with pixels having the same sequence number in successive paths. And numerically flagging those path pixels found in the comparison to have a predetermined intensity relationship to the centroid pixel, and having the same order in paths tested sequentially. Clutter rejection of the digitized image, which comprises the steps of: flagging all pixels with a flag, and identifying all potential target objects as true if all pixels in the nested path are flagged. Method. 6. The sequence number is assigned to the path number of a pair, and a pair of second numbers are opposite the diameter of the pixel to which the first number of the pair is assigned with respect to the centroid pixel. The clutter elimination method according to claim 4, wherein the clutter elimination method is assigned to a pixel located. 7. Finding a potential target centroid pixel in the image, all pixels of the entire path having a predetermined intensity relationship with respect to the centroid pixel, and all pixels of the entire path from the centroid pixel. Determining the existence of a closed path of pixels surrounding the centroid pixel so that it is located within a predetermined distance, and identifying such target object as true when such path is determined to exist. A method for eliminating clutter in a digitized image, the method comprising the steps of: 8. The determining step defines a plurality of nested closed pixel paths surrounding the centroid, and determines the existence of the predetermined intensity relationship for successive pixels within one of the paths. Then, for each pixel of the internal path for which the relationship does not exist, the relationship exists for a pixel located radially outward on the next closed path radially outward of the closed path. The presence of a closed true path of pixels having the predetermined relationship can be determined even if the true path has a different shape than the nested path. The method for eliminating clutter of a digitized image according to claim 7.