System and method for detecting and tracking objects

Abstract

This disclosure relates to systems and methods for detecting and tracking objects. One method includes: receiving a first image captured at a first time. The method further includes: detecting the position of a first object in the first image. The method further includes: determining a region of interest based at least in part on the position of the first object in the first image. The method further includes: receiving a second image captured at a second time. The method further includes: identifying the region of interest in the second image. The method further includes: detecting the position of a second object in the second image located outside the region of interest.

Description

Technical Field

[0001] This disclosure relates to tracking an aircraft. More specifically, this disclosure relates to detecting a first object (e.g., an aircraft) over a long distance, tracking the first object after detection, and then detecting a second object while tracking the first object. Background Technology

[0002] Methods exist for detecting single-pixel or subpixel objects in image or video frames corresponding to moving aircraft over long distances. Because aircraft are unlikely to be identifiable in every image or video frame, these methods are called dark object detection. Therefore, the detection problem is more difficult than that solved by conventional object detection neural networks, and the subsequent tracking problem (i.e., tracking the detected object) is more challenging than that solved by conventional object trackers.

[0003] A conventional long-range air-to-air vehicle detector uses a Hidden Markov Model (HMM) for morphological image processing and pre-detection tracking. However, the HMM model used by this detector focuses on a single object (e.g., an aircraft) and therefore cannot detect more than one object at a time. Furthermore, once an object is detected in an image or video frame, the detector can switch back to a state with a low detection probability. This state occurs when the object moves away or due to noise in the image or video frame. Summary of the Invention

[0004] A method is disclosed. The method includes: identifying a first image captured at a first time. The method further includes: detecting the position of a first object in the first image. The method further includes: determining a region of interest based at least in part on the position of the first object in the first image. The method further includes: identifying a second image captured at a second time. The method further includes: identifying the region of interest in the second image. The method further includes: detecting the position of a second object in the second image located outside the region of interest.

[0005] A computing system is disclosed. The computing system includes one or more processors and a memory system. The memory system includes one or more non-volatile computer-readable media storing instructions that, when executed by at least one of the one or more processors, cause the computing system to perform operations. The operations include: recognizing a first image. The first image is captured by a camera on a first aircraft at a first time. The operations further include: detecting the position of a first object in the first image. The first object is moving. The operations further include: determining a region of interest based at least in part on the position of the first object in the first image. The region of interest includes a plurality of pixels in the first image. Each pixel in the region of interest has a probability greater than the predetermined threshold of the first object being located therein at a second time. The operations further include: recognizing a second image. The second image is captured by a camera on a first aircraft at a second time. The operations further include: recognizing the region of interest in the second image. The operations further include: detecting the position of a second object in the second image located outside the region of interest, wherein the second object is moving.

[0006] In another embodiment, the operation includes: identifying a first image. The first image is captured by a camera on a first aircraft at a first time. The operation further includes: detecting the position of a first object in the first image using a Hidden Markov Model. The first object is a second aircraft in flight. The first object is represented as five or fewer pixels in the first image. The operation further includes: determining a region of interest (ROI) at least partially based on the position of the first object in the first image. The first object is located within the ROI. The ROI includes multiple pixels in the first image. Each pixel in the ROI has a probability greater than a predetermined threshold of the first object being located therein at a second time. The second time is after the first time. The operation further includes: identifying a second image. The second image is captured by a camera on the first aircraft at a second time. The operation further includes: identifying the ROI in the second image. The operation further includes: detecting the position of the first object in the ROI of the second image using a Hidden Markov Model. The operation further includes: detecting the position of the second object in the second image located outside the ROI using a Hidden Markov Model to avoid re-detecting the first object. The second object is a third aircraft in flight. The operation further includes: predicting the trajectories of the first object and the second object based at least in part on the position of the first object in the first image, the position of the first object in the second image, and the position of the second object in the second image. The operation also includes: navigating the first aircraft based at least in part on the trajectories of the first object and the second object. Attached Figure Description

[0007] The accompanying drawings, which are incorporated into and form part of this specification, illustrate various aspects of this teaching and, together with the specific embodiments, serve to explain the principles of this teaching.

[0008] Figure 1 A schematic diagram of a first aircraft in flight according to an embodiment is shown.

[0009] Figure 2 A flowchart of a method for detecting and tracking one or more objects according to an embodiment is shown.

[0010] Figure 3 A first image captured by a camera on a first aircraft is shown according to an embodiment.

[0011] Figure 4 A first image is shown, which, according to an embodiment, contains an identified first object (e.g., a second aircraft).

[0012] Figure 5 A first image showing a first region of interest identified around a first object, according to an embodiment, is shown.

[0013] Figure 6 A second image captured by a camera on a first aircraft is shown according to an embodiment.

[0014] Figure 7 A second image, which contains the identified first object, is shown according to an embodiment.

[0015] Figure 8 A second image is shown, which, according to an embodiment, contains an identified second object (e.g., a third aircraft).

[0016] It should be noted that some details in the diagram have been simplified and drawn for ease of understanding, rather than maintaining strict structural accuracy, detail, and proportion. Detailed Implementation

[0017] Embodiments thereof are now shown in detail with reference to the present teachings and in the accompanying drawings. In the drawings, similar reference numerals are used throughout to denote the same elements. In the following description, reference is made to the accompanying drawings, which form a part of the teachings, and which are shown by way of illustrative specific embodiments of the implementation of the present teachings. Therefore, the following description is merely exemplary.

[0018] The system and method disclosed herein create an object tracker capable of (1) tracking detected objects after initialization by a detector, and (2) setting the detection probability within the region of the tracked object to 0 for a detector moving forward. This second part does not use a pre-detection tracking model that typically focuses only on one detection, such as a Hidden Markov Model (HMM), to continue searching for more objects within the frame. This is achieved by accepting a list of 2D points corresponding to detected points passed from a morphological HMM detector as input “tracking_points” and accepting a 1-channel image where each pixel value corresponds to the probability that an aircraft is present at that location as “detection”.

[0019] Systems and methods can be used to detect objects in the air (e.g., in flight). When objects such as birds or projectiles are uncooperative, systems and methods can be used to sense and avoid them.

[0020] Figure 1 A schematic diagram of an aircraft 100 in flight according to an embodiment is shown. The aircraft 100 may be or include an airplane, helicopter, unmanned aerial vehicle (e.g., drone or satellite), spacecraft, etc. The aircraft 100 may include a camera 110. The camera 110 may be coupled to and / or positioned within the aircraft 100. The camera 110 may be configured to capture one or more images. The camera 110 may also be configured, or instead, to capture a continuous stream of images (i.e., video) over time. The camera 110 may be configured to capture video, and the images may be still frames extracted from the video. In another embodiment, the camera 110 may not be mounted on the aircraft 100. Instead, the camera 110 may be located on or coupled to a building such as a control tower. The camera 110 may be mounted to a directional support operable to adjust azimuth and / or attitude.

[0021] Images and / or video can be transmitted to a computing system 120 located on aircraft 100. In another embodiment, the computing system 120 may be located on the ground (e.g., in a control station) in communication with the onboard computing system. The computing system may be or include one or more processors and a memory system. The memory system includes one or more non-volatile computer-readable media storing instructions that, when executed by at least one of the one or more processors, cause the computing system to perform operations. Reference is made below. Figure 2 Describe the operation.

[0022] Figure 2A flowchart of a method 200 for detecting and tracking one or more objects according to an embodiment is shown. More specifically, method 200 may detect a first object over a long distance, track the first object after detection, and then detect a second object while tracking the first object. The illustrated order of method 200 is described below. One or more steps in method 200 may be performed, repeated, or omitted in a different order.

[0023] Method 200 may include: at 202, receiving or recognizing a first image 300. Figure 3 An embodiment of a first image 300 is illustrated. The first image 300 can be captured by camera 110 and can be sent to / received by computing system 120. Camera 110 can be positioned facing the direction of flight such that the first image 300 is located in front of aircraft 100. Thus, the first image 300 can include at least a portion of the flight path of aircraft 100. The first image 300 can include non-sky areas such as ground area 310, sky area 320, or both. In at least one embodiment, the first image 300 can be color and can be converted to grayscale and / or binary (e.g., black and white). In another embodiment, the first image 300 can be grayscale and / or binary (e.g., black and white).

[0024] Method 200 may further include: at 204, detecting the position of the first object 410 in the first image 300. This step can be performed by the computing system 120. Figure 4 The location of the first object 410 is identified. The first object 410 can be detected using a Hidden Markov Model (HMM). More specifically, the first object 410 can be detected using image processing techniques such as morphological on / off.

[0025] The first object 410 may be or include a second aircraft in flight. In another embodiment, the first object 410 may be one or more flying creatures (e.g., a flock of birds, a swarm of insects, etc.). The first object 410 may be represented by only a small number of pixels in the first image 300. For example, the first object 410 may be represented by fewer than 50 pixels, fewer than 30 pixels, fewer than 10 pixels, fewer than 5 pixels, or by a single pixel.

[0026] Method 200 may further include: at 206, determining a region of interest 510 in the first image 300. This step can be performed by the computing system 120. The region of interest 510 is identified in... Figure 5The region of interest 510 can be determined at least in part based on the position of the first object 410 in the first image 300. The region of interest 510 can include more pixels than the first object 410. For example, the region of interest 510 can include 2 to 5 pixels, 5 to 10 pixels, 10 to 20 pixels, 20 to 50 pixels, or 50 to 100 pixels. The region of interest 510 can include about 1% to about 2% of the pixels in the first image 300, about 2% to about 4% of the pixels in the first image 300, about 4% to about 6% of the pixels in the first image 300, or about 6% to about 10% of the pixels in the first image 300. The region of interest 510 can include the first object 410. In other words, the first object 410 can be located within the region of interest 510. In the illustrated embodiment, the region of interest 510 can be generally circular, and the first object 410 can be located at the center of the circle.

[0027] As discussed below, each pixel in the region of interest 510 may have a probability greater than a predetermined threshold in which at least a portion of the first object 410 is located at a second time. Pixels in the first image 300 located outside the region of interest 510 may have a probability less than the predetermined threshold in which the first object 410 is located at a second time. As described below, when determining the next position of the first object 410 (e.g., in the second image 600), pixels located outside the region of interest 510 may be disregarded / analyzed.

[0028] In one embodiment, the computing system 120 may include a machine learning (ML) probabilistic model trained to determine the probability that a pixel in the region of interest 510 comprises part of a first object 410 at a second time. The threshold may vary depending on temporal variations between captured images, image resolution, and / or the category of the detected object (e.g., size, shape, etc.).

[0029] Method 200 may further include: at 208, receiving or recognizing a second image 600. Figure 6 An embodiment of a second image 600 is illustrated. The second image 600 may be captured by camera 110 and transmitted to / received by computing system 120. Camera 110 may be positioned such that the second image 600 is located in front of aircraft 100. Thus, the second image 600 may include at least a portion of the flight path of aircraft 100. The second image 600 may include a non-sky region 310, a sky region 320, or both. In at least one embodiment, the second image 600 may be color and may be converted to grayscale and / or binary (e.g., black and white). In another embodiment, the second image 600 may be grayscale and / or binary (e.g., black and white).

[0030] The second image 600 can be captured after the first image 300. For example, the second image 600 can be captured within about 0.1 seconds to about 1 second, about 1 second to about 5 seconds, about 5 seconds to about 30 seconds, or about more than 30 seconds after the first image 300. Thus, the second image 600 may also include the first object 410.

[0031] Method 200 may further include: at 210, identifying a region of interest 510 in the second image 600. This in Figure 7 As shown in the figure. The region of interest 510 can represent the same pixels in the first image 300 and the second image 600. For example, the center of the region of interest 510 can be 100 pixels to the right (from the left side of the image) and 50 pixels down (from the top of the image) in both images 300 and 600.

[0032] Method 200 may further include: at 212, detecting the position of the first object 410 in the second image 600. This step can be performed by the computing system 120. Figure 7 The first object 410 is identified in the image. Detection of the first object 410 can occur only within the region of interest 510 of the second image 600. The location of the first object 410 in the second image 600 can be detected without analyzing pixels outside the region of interest 510. This is computationally more efficient than attempting to detect the first object 410 across the entire second image 600. The first object 410 can be detected using a Hidden Markov Model (HMM).

[0033] In the second image 600, the first object 410 may be represented by at least one pixel or only a few pixels. For example, the first object 410 may be represented by fewer than 50 pixels, fewer than 30 pixels, fewer than 10 pixels, fewer than 5 pixels, or by a single pixel. The bottom of the arrow 710 indicates the position of the first object 410 in the first image 300, and the top of the arrow 710 indicates the position of the first object 410 in the second image 600. Thus, the arrow 710 indicates the distance the first object 410 has moved from the first image 300 to the second image 600. In other words, the arrow 710 indicates the distance the first object 410 has moved from a first time (e.g., when the first image 300 is captured) to a second time (e.g., when the second image 600 is captured).

[0034] Method 200 may further include: at 214, detecting the position of the second object 810 in the second image 600. This step can be performed by the computing system 120. Figure 8 The location of the second object 810 is marked in the middle. The second object 810 can also be detected using an HMM.

[0035] The second object 810 can be detected in a portion of the second image 600 outside the region of interest 510 to avoid re-detecting the first object 410. More specifically, the HMM object tracker can only track one object at a time. Method 200 allows tracking of the first object 410 while detecting and / or tracking the second object 810. The second object 810 may be different from the first object 410. For example, the second object 810 may be or include a third flying object in flight. In another embodiment, the second object 810 may be one or more flying organisms (e.g., a flock of birds, a swarm of insects, etc.).

[0036] Method 200 may further include, at 216, predicting the route and / or trajectory of the first object 410 and the second object 810. This step may be performed by the computing system 120. The route may be or include a series of waypoints describing the path that the aircraft 100 will take. The path may be or include the places where the aircraft 100 is flying and will fly during the mission. The trajectory may be or include a series of aircraft positions and speeds representing the aircraft's path relative to time. For example, the trajectory may be a route with an additional time component inserted. The trajectory of the first object 410 may be predicted at least in part based on the position of the first object 410 in the first image 300, its position in the second image 600, or both. The trajectory of the second object 810 may be predicted at least in part based on the position of the second object 810 in the second image 600.

[0037] Method 200 may further include, at 218, navigating (e.g., turning) the first aircraft 100. Navigation (e.g., turning) of the aircraft 100 may be based at least in part on a first image 300, the position of the first object 410 in the first image 300, a second image 600, the position of the first object 410 in the second image 600, the position of the second object 810 in the second image 600, a predicted trajectory of the first object 410, a predicted trajectory of the second object 810, or a combination thereof. For example, the aircraft 100 may be navigated (e.g., turned) to avoid collision paths with the first object 410 and / or the second object 810. Navigation may be performed (e.g., automatically) by a computing system 120. In another embodiment, navigation may be performed by a user. The user may be inside the aircraft 100 (e.g., a pilot), or the user may be on the ground and operate the aircraft 100 remotely.

[0038] Method 200 may be iterative. For example, method 200 may further include: determining a first region of interest and a second region of interest in the second image 600. The first region of interest may include a first object 410, and the second region of interest may include a second object. The first region of interest in the second image 600 may be different from the region of interest 510 in the second image 600. More specifically, the first object 410 may be located at or near the center of the first region of interest in the second image 600. The first region of interest in the second image 600 may at least partially overlap with the region of interest 510 in the second image 600.

[0039] Then, method 200 may include: receiving a third image captured after the first image 300 and the second image 600. Then, method 200 may include: identifying a first region of interest and a second region of interest in the third image. Then, method 200 may include: detecting the positions of a first object 410 and a second object 810 in the third image. Detection of the first object 410 may occur only within the first region of interest in the third image, and detection of the second object 810 may occur only within the second region of interest in the third image. Then, method 200 may include: detecting the position of a third object in the third image. The third object may be detected in a portion of the third image located outside the first and second regions of interest 510 to avoid re-detection of the first object 410 and the second object 810. The third object 810 may be different from the first object 410 and the second object 810. For example, the third object may be or include a fourth aircraft in flight. In another embodiment, the third object may be one or more flying organisms (e.g., a flock of birds, a swarm of insects, etc.).

[0040] Then, method 200 may include: predicting the trajectories of the first object 410, the second object 810, and the third object. For example, this may include: updating the predicted trajectory of the first object 410 at least partially based on the first image 300, the position of the first object 410 in the first image 300, the second image 600, the position of the first object 410 in the second image 600, the third image, the position of the first object in the third image, the predicted trajectory of the first object 410 (from step 216), or a combination thereof. This may also include: updating the predicted trajectory of the second object 810 at least partially based on the second image 600, the position of the second object 810 in the second image 600, the third image, the position of the second object 810 in the third image, the predicted trajectory of the second object 810 (from step 216), or a combination thereof. This may also include: predicting the trajectory of the third object at least partially based on the third image, the position of the third object in the third image, or a combination thereof. Then, method 200 may include: adjusting the navigation (e.g., maneuvering) of the first aircraft 100 at least partially based on the predicted trajectories. In another embodiment, method 200 may instead include: adjusting camera 110 using a directional support.

[0041] As used herein, the terms “inner” and “outer,” “upper” and “lower,” “above” and “below,” “upward” and “downward,” “upstream” and “downstream,” “above” and “below,” “inward” and “outward,” etc., refer to their relative positions and are not intended to indicate specific directions or spatial orientations. The terms “couple,” “coupled,” “connect,” “connection,” “connected,” “in connectionwith,” and “connecting” refer to “direct connection” or “connection via one or more intermediate elements or components.” Similarly, the terms “bonded” and “bonding” refer to “direct bonding” or “bonding via one or more intermediate elements, components, or layers.”

[0042] Although the numerical ranges and parameters described in this disclosure are approximate, the values ​​described in the specific embodiments are reported as precisely as possible. However, any numerical value inherently includes a particular error that is necessarily arising from the standard deviation found in its corresponding test measurement. Moreover, all ranges disclosed herein should be understood to include any and all subranges summarized herein.

[0043] Although the teachings have been shown with respect to one or more embodiments, changes and / or modifications can be made to the illustrated embodiments without departing from the spirit and scope of the appended claims. Furthermore, although specific features of the teachings have been disclosed with respect to only one embodiment among several embodiments, such features may be combined with one or more other features in other embodiments as desired and advantageous for any given specific function. As used herein, the terms “a,” “an,” and “the” may refer to one or more elements or portions of elements. As used herein, the terms “first” and “second” may refer to two distinct elements or portions of elements. As used herein, the term “at least one of A and B” relative to a list of items such as, for example, A and B, refers to A alone, B alone, or A and B. Those skilled in the art will recognize that these and other variations are possible. Further, the use of the terms “including,” “includes,” “having,” “has,” “with,” or variations thereof in the detailed description and claims is intended to include the term in a manner similar to the term “comprising.” Furthermore, in the discussion and claims herein, the term "approximately" means that the listed values ​​may be changed to a certain extent, provided that the changes do not cause the process or structure to be inconsistent with the intended purpose described herein. Finally, "exemplary" means that the description is used as an example and not implying that it is idealized.

[0044] It should be recognized that the variations and other features and functions disclosed above, or their alternatives, can be combined into a variety of other different systems or applications. Those skilled in the art can then make various unforeseen or unanticipated alternatives, modifications, variations, or improvements, which are also intended to be covered by the following claims.

[0045] Item 1: A method comprising: identifying a first image captured at a first time; detecting the location of a first object in the first image; determining a region of interest based at least in part on the location of the first object in the first image; identifying a second image captured at a second time; identifying the region of interest in the second image; and detecting the location of a second object in a portion of the second image located outside the region of interest.

[0046] Item 2: The method according to Item 1, wherein the first image is captured by a camera on the aircraft.

[0047] Item 3: The method according to Item 1 or 2, wherein the first object is an aircraft in flight.

[0048] Item: The method according to any one of items 1 to 3, wherein the first object is represented at least as a pixel in the first image.

[0049] Item 5: The method according to any one of items 1 to 4, wherein the region of interest includes a plurality of pixels in the first image, and wherein the first object is located within the plurality of pixels.

[0050] Item 6: The method according to any one of items 1 to 5, wherein each pixel in the region of interest has a probability greater than a predetermined threshold in which the first object is located at the second time, wherein each pixel outside the region of interest has a probability less than the predetermined threshold in which the first object is located at the second time, and wherein the second time is after the first time.

[0051] Item 7: The method according to any one of items 1 to 6, wherein the region of interest comprises a circle, and wherein the first object is located at the center of the circle in the first image.

[0052] Item 8: The method according to any one of items 1 to 7 further includes: detecting the position of the first object in the region of interest of the second image, wherein the first object is not located at the center of a circle in the second image.

[0053] Item 9: The method according to Item 8 further includes: predicting the trajectory of the first object based at least in part on the position of the first object in the first image and the second image.

[0054] Item 10: The method according to item 10 further includes: navigating the aircraft based at least in part on the trajectory of the first object.

[0055] Item 11: A computing system comprising: one or more processors; and a memory system including one or more non-volatile computer-readable media storing instructions that, when executed by at least one of the one or more processors, cause the computing system to perform operations including: identifying a first image, wherein the first image is captured by a camera on a first aircraft at a first time; detecting the position of a first object in the first image, wherein the first object is moving; determining a region of interest based at least in part on the position of the first object in the first image, wherein the region of interest includes a plurality of pixels in the first image, and wherein each pixel in the region of interest has a probability greater than the predetermined threshold of the first object being located therein at a second time; identifying a second image, wherein the second image is captured by a camera on a first aircraft at a second time; identifying the region of interest in the second image; and detecting the position of a second object in the second image, located outside the region of interest, wherein the second object is moving.

[0056] Item 12: The computing system according to Item 11, wherein the position of the first object in the first image is detected using a hidden Markov model that analyzes the entire first image.

[0057] Item 13: The computing system according to Item 12, wherein the position of the second object in the second image is detected using a hidden Markov model that analyzes only the portion of the second image located outside the region of interest.

[0058] Item 14: The computing system according to Item 13 further includes: detecting the position of the first object in a region of interest in the second image, wherein the position of the first object in the second image is detected using a hidden Markov model that analyzes only the region of interest in the second image.

[0059] Item 15: The computing system according to Item 14 further includes: predicting the trajectories of the first object and the second object based at least in part on the position of the first object in the first image, the position of the first object in the second image, and the position of the second object in the second image; and navigating the first aircraft based at least in part on the trajectories of the first object and the second object.

[0060] Item 16: A computing system comprising: one or more processors; and a memory system including one or more non-volatile computer-readable media storing instructions, which, when executed by at least one of the one or more processors, cause the computing system to perform operations including: recognizing a first image, wherein the first image is captured by a camera on a first aircraft in a first instant; detecting the position of a first object in the first image using a hidden Markov model, wherein the first object is a second aircraft in flight, and wherein the first object is represented as five or fewer pixels in the first image; determining a region of interest based at least in part on the position of the first object in the first image, wherein the first object is located within the region of interest, wherein the region of interest comprises a plurality of pixels in the first image, wherein each pixel in the region of interest has a value greater than a first... The probability that an object is located within a predetermined threshold at a second time, wherein the second time is after the first time; identifying a second image, wherein the second image is captured by a camera on the first aircraft at the second time; identifying a region of interest in the second image; detecting the position of the first object in the region of interest in the second image using a Hidden Markov Model; detecting the position of the second object in the second image outside the region of interest using a Hidden Markov Model to avoid re-detecting the first object, wherein the second object is a third aircraft in flight; predicting the trajectories of the first object and the second object based at least in part on the position of the first object in the first image, the position of the first object in the second image, and the position of the second object in the second image; and navigating the first aircraft based at least in part on the trajectories of the first object and the second object.

[0061] Item 17: The computing system according to Item 16 further includes: determining a first region of interest in the second image based at least in part on the position of a first object in the second image, wherein the first object is located at the center of the first region of interest in the second image; and determining a second region of interest in the second image based at least in part on the position of a second object in the second image, wherein the second object is located at the center of the second region of interest in the second image.

[0062] Item 18: The computing system according to Item 17 further includes: identifying a third image, wherein the third image is captured by a camera on the first aircraft at a third time; identifying a first region of interest and a second region of interest in the third image; detecting the position of a first object in the first region of interest of the third image; and detecting the position of a second object in the second region of interest of the third image.

[0063] Item 19: The computing system according to Item 18 further includes: detecting the position of a third object in a portion of the third image located outside the first region of interest and the second region of interest, so as to avoid re-detecting the first object and the second object.

[0064] Item 20: The computing system according to Item 19 further includes: navigating the first aircraft based at least in part on the position or trajectory of a third object.

Claims

1. A method (200) for detecting and tracking objects, comprising: Identify the first image (300) captured by the camera in the first instant; Detect the position of the first moving object (410) in the first image (300); A region of interest (510) is determined at least in part based on the position of the first moving object (410) in the first image (300), wherein the region of interest includes a plurality of pixels in the first image, and wherein each pixel in the region of interest has a probability greater than a predetermined threshold in which the first moving object is located at a second time. Identify the second image (600) captured by the camera at the second time. Identify the region of interest (510) in the second image (600); and The position of a second moving object (810) is detected in the portion of the second image (600) located outside the region of interest (510).

2. The method according to claim 1, wherein, The first image (300) is captured by a camera (110) on the aircraft (100).

3. The method according to claim 1, wherein, The first moving object (410) is a flying aircraft.

4. The method according to claim 1, wherein, The first moving object (410) is represented at least as a pixel in the first image (300).

5. The method according to claim 1, wherein, And among them, The first moving object (410) is located within the plurality of pixels.

6. The method according to claim 1, wherein, Each pixel located outside the region of interest (510) has a probability less than a predetermined threshold in which the first moving object (410) is located at the second time, and wherein the second time is after the first time.

7. The method according to claim 1, wherein, The region of interest (510) comprises a circle, wherein the first moving object (410) is located at the center of the circle in the first image (300).

8. The method of claim 7, further comprising: The position of the first moving object (410) in the region of interest (510) of the second image (600) is detected, wherein the first moving object (410) is not located at the center of the circle in the second image (600). The method further includes: predicting the trajectory of the first moving object (410) based at least in part on the position of the first moving object (410) in the first image (300) and the second image (600); and The method further includes navigating the aircraft (100) based at least in part on the trajectory of the first moving object (410).

9. A computing system (120), comprising: One or more processors; and A memory system including one or more non-volatile computer-readable media storing instructions that, when executed by at least one of the one or more processors, cause the computing system to perform operations, the operations including: Identify the first image (300), wherein the first image (300) is captured by a camera (110) on the first aircraft (100) in a first instant; Detect the position of the first object (410) in the first image (300), wherein the first object (410) is moving; A region of interest (510) is determined at least in part based on the position of the first object (410) in the first image (300), wherein the region of interest (510) includes a plurality of pixels in the first image (300), and wherein each pixel in the region of interest (510) has a probability greater than a predetermined threshold in which the first object (410) is located at a second time. Identify a second image (600), wherein the second image (600) is captured by the camera (110) on the first aircraft (100) at the second time; Identify the region of interest (510) in the second image (600); and The location of a second object (810) in the second image (600) outside the region of interest (510) is detected, wherein the second object (810) is moving.

10. The computing system according to claim 9, wherein, The position of the first object (410) in the first image (300) is detected using a hidden Markov model that analyzes the entire first image (300).

11. The computing system according to claim 10, wherein, The location of the second object (810) in the second image (600) is detected using the Hidden Markov Model, which analyzes only the portion of the second image (600) located outside the region of interest (510).

12. The computing system according to claim 11, further comprising: The position of the first object (410) in the region of interest (510) of the second image (600) is detected, wherein the position of the first object (410) in the second image (600) is detected using the Hidden Markov Model that analyzes only the region of interest (510) in the second image (600).

13. The computing system according to claim 12, further comprising: The trajectories of the first object (410) and the second object (810) are predicted at least in part based on the position of the first object (410) in the first image (300), the position of the first object (410) in the second image (600), and the position of the second object (810) in the second image (600); and The first aircraft (100) is navigated based at least in part on the trajectories of the first object (410) and the second object (810).

14. A computing system (120), comprising: One or more processors; and A memory system including one or more non-volatile computer-readable media storing instructions that, when executed by at least one of the one or more processors, cause the computing system to perform operations, the operations including: Identify the first image (300), wherein the first image (300) is captured by a camera (110) on the first aircraft (100) in a first instant; The position of a first object (410) in the first image (300) is detected using a hidden Markov model, wherein the first object (410) is a second aircraft in flight, and wherein the first object (410) is represented by five or fewer pixels in the first image (300). A region of interest (510) is determined at least in part based on the position of the first object (410) in the first image (300), wherein the first object (410) is located within the region of interest (510), wherein the region of interest (510) comprises a plurality of pixels in the first image (300), wherein each pixel in the region of interest (510) has a probability greater than a predetermined threshold in which the first object (410) is located at a second time, and wherein the second time is after the first time; Identify a second image (600), wherein the second image (600) is captured by the camera (110) on the first aircraft (100) at the second time; Identify the region of interest (510) in the second image (600); The location of the first object (410) in the region of interest (510) of the second image (600) is detected using the hidden Markov model; The hidden Markov model is used to detect the location of the second object (810) in the portion of the second image (600) outside the region of interest (510) to avoid re-detecting the first object (410), wherein the second object (810) is a third aircraft in flight; The trajectories of the first object (410) and the second object (810) are predicted at least in part based on the position of the first object (410) in the first image (300), the position of the first object (410) in the second image (600), and the position of the second object (810) in the second image (600); and The first aircraft (100) is navigated based at least in part on the trajectories of the first object (410) and the second object (810).

15. The computing system of claim 14, further comprising: A first region of interest in the second image is determined at least in part based on the position of the first object (410) in the second image (600), wherein the first object (410) is located at the center of the first region of interest in the second image (600); and The second region of interest in the second image (600) is determined at least in part based on the position of the second object (810) in the second image (600), wherein the second object (810) is located at the center of the second region of interest in the second image (600).

16. The computing system according to claim 15, further comprising: Identify a third image, wherein the third image is captured by the camera (110) on the first aircraft (100) at a third time. Identify the first region of interest and the second region of interest in the third image; Detect the position of the first object (410) in the first region of interest of the third image; and Detect the position of the second object (810) in the second region of interest of the third image.

17. The computing system of claim 16, further comprising: The location of the third object in the third image outside the first region of interest and the second region of interest is detected to avoid re-detection of the first object (410) and the second object (810).

18. The computing system according to claim 17, further comprising: The first aircraft (100) is navigated based at least in part on the position or trajectory of the third object.

Images