CN109960990B - Method for evaluating reliability of obstacle detection - Google Patents
Method for evaluating reliability of obstacle detection Download PDFInfo
- Publication number
- CN109960990B CN109960990B CN201711431445.5A CN201711431445A CN109960990B CN 109960990 B CN109960990 B CN 109960990B CN 201711431445 A CN201711431445 A CN 201711431445A CN 109960990 B CN109960990 B CN 109960990B
- Authority
- CN
- China
- Prior art keywords
- obstacle
- different
- current
- detection result
- obstacle detection
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
- G06V20/58—Recognition of moving objects or obstacles, e.g. vehicles or pedestrians; Recognition of traffic objects, e.g. traffic signs, traffic lights or roads
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Image Analysis (AREA)
Abstract
An obstacle detection reliability evaluation method implemented by a processing unit includes: (A) after receiving a current image, obtaining a first current obstacle detection result indicating an obstacle detected by a first classification method by using the first classification method, and obtaining a second current obstacle detection result indicating an obstacle detected by a second classification method by using the second classification method; (B) determining whether the first and second current obstacle detection results have at least one different obstacle; and (C) when it is determined that the at least one different obstacle exists in the first current obstacle detection result and the second current obstacle detection result, obtaining at least one penalty score corresponding to the at least one different obstacle, respectively, and subtracting the at least one penalty score from the confidence score.
Description
Technical Field
The present invention relates to image data processing, and more particularly, to a method for evaluating reliability of obstacle detection.
Background
With the popularization of cameras and the development of computer vision field, the application of intelligent image monitoring brings safety and convenience to human life, for example, the application is applied to an intelligent Advanced Driver Assistance System (ADAS) to detect obstacles on an image shot by a camera by using an image recognition technology so as to warn a dangerous road condition for driving and decelerate or stop a vehicle in some cases.
However, the actual road belongs to a complex environment, and is easily affected by the intervention of external environmental factors, and when the camera is affected by the intervention of external environmental factors (for example, the direct light of the vehicle facing the lane, the backlight or the pollution of the camera lens), the texture and the contour of the obstacle in the captured image are also affected, thereby causing the detection of the obstacle to be misaligned. However, the conventional obstacle detection method of the ADAS system only outputs the detection result of the obstacle and does not provide the confidence level of the detection result, which may cause the user to misunderstand that the detection result of the ADAS system is extremely accurate and excessively believes and depends on the detection result of the ADAS system.
Disclosure of Invention
The invention aims to provide an evaluation method capable of timely providing the reliability of obstacle detection.
The obstacle detection reliability evaluation method is implemented by a processing unit electrically connected with a storage unit and an image shooting unit, the storage unit stores a reliability score, the image shooting unit continuously shoots and transmits a series of images to the processing unit, and the obstacle detection reliability evaluation method comprises a step (A), a step (B) and a step (C).
In the step (a), after the processing unit receives the current image from the image capturing unit, the processing unit obtains and stores a first current obstacle detection result indicating an obstacle detected by the first classification method by using a first classification method according to the current image, and obtains and stores a second current obstacle detection result indicating an obstacle detected by the second classification method by using a second classification method according to the current image.
In step (B), the processing unit determines whether at least one different obstacle exists in the first current obstacle detection result and the second current obstacle detection result.
In the step (C), when the processing unit determines that the first current obstacle detection result and the second current obstacle detection result have the at least one different obstacle, the processing unit obtains at least one penalty score corresponding to the at least one different obstacle respectively according to the at least one different obstacle, and subtracts the at least one penalty score from the confidence score to update the confidence score.
Preferably, the obstacle detection reliability evaluation method of the present invention, step (C) includes the following substeps:
(C-1) when it is determined that the first current obstacle detection result and the second current obstacle detection result have the at least one different obstacle, determining, for each of the at least one different obstacle, whether the first current obstacle detection result does not have the different obstacle;
(C-2) when it is determined that the different obstacle does not exist as the first current obstacle detection result, obtaining a first penalty score, which is one of the at least one penalty score, according to at least the different obstacle, and subtracting the first penalty score from the confidence score to update the confidence score; and
(C-3) when it is determined that the different obstacle exists in the first current obstacle detection result, obtaining a second penalty score, which is one of the at least one penalty score, according to at least the different obstacle, and subtracting the second penalty score from the confidence score to update the confidence score.
Preferably, in the obstacle detection reliability evaluation method according to the present invention, the storage unit further stores a first lookup table of obstacle pair scores, a first penalty weight corresponding to the first classification method, and a second penalty weight corresponding to the second classification method, the first lookup table includes a plurality of obstacle distances related to distances between actual obstacles and the image capturing unit, and a plurality of scores respectively corresponding to the obstacle distances:
in step (a), the first current obstacle detection result and the second current obstacle detection result both include the detected obstacle, and the detected obstacle is located at the position of the current image;
in step (C-2), the processing unit estimates a distance between an actual obstacle corresponding to the different obstacle and the image capturing unit according to the position of the different obstacle in the current image, obtains a score corresponding to the different obstacle according to the estimated distance and the first lookup table, and multiplies the obtained score by the first penalty weight to obtain the first penalty score; and
in the step (C-3), the processing unit first estimates a distance between an actual obstacle corresponding to the different obstacle and the image capturing unit according to a position of the different obstacle in the current image, obtains a score corresponding to the different obstacle according to the estimated distance and the first lookup table, and multiplies the obtained score by the second penalty weight to obtain the second penalty score.
Preferably, in the obstacle detection reliability evaluation method of the present invention, the storage unit further stores a first lookup table of obstacle pair scores, a first penalty weight corresponding to the first classification method, and a second penalty weight corresponding to the second classification method, the first lookup table includes a plurality of obstacle categories, and a plurality of scores respectively corresponding to the categories:
in the step (a), the first current obstacle detection result and the second current obstacle detection result both include the detected obstacle and the type of the detected obstacle;
in step (C-2), the processing unit obtains the scores corresponding to the different obstacles according to the types of the different obstacles and the first lookup table, and multiplies the obtained scores by the first penalty weight to obtain the first penalty score; and
in step (C-3), the processing unit obtains the scores corresponding to the different obstacles according to the types of the different obstacles and the first lookup table, and multiplies the obtained scores by the second penalty weight to obtain the second penalty score.
Preferably, after the step (a), the obstacle detection reliability evaluation method of the present invention further includes the following steps:
(D) determining whether at least one different obstacle exists in a first previous obstacle detection result and a first current obstacle detection result, which are obtained according to a previous image and indicate the obstacle detected by the first classification method;
(E) when the first previous obstacle detection result and the first current obstacle detection result are judged to have the at least one different obstacle, at least one third penalty score corresponding to the at least one different obstacle is obtained according to the at least one different obstacle, and the confidence score is subtracted by the at least one third penalty score so as to update the confidence score.
Preferably, in the obstacle detection reliability evaluation method according to the present invention, the storage unit further stores a second lookup table of obstacle pair scores, the second lookup table includes a plurality of obstacle positions related to positions of obstacles in the image, and a plurality of scores respectively corresponding to the obstacle positions:
in step (a), the first current obstacle detection result and the second current obstacle detection result both include the detected obstacle, and the detected obstacle is located at the position of the current image; and
in step (E), the processing unit obtains the scores corresponding to the different obstacles according to the positions of the different obstacles in the current image or the previous image and the second lookup table, and obtains the third penalty score according to the obtained scores.
Preferably, after the step (a), the obstacle detection reliability evaluation method of the present invention further includes the following steps:
(F) determining whether at least one different obstacle exists in a second previous obstacle detection result and a second current obstacle detection result, which are obtained according to a previous image and indicate the obstacle detected by the second classification method;
(G) when the second previous obstacle detection result and the second current obstacle detection result are judged to have the at least one different obstacle, at least one fourth penalty score corresponding to the at least one different obstacle is obtained according to the at least one different obstacle, and the confidence score is subtracted by the at least one fourth penalty score so as to update the confidence score.
Preferably, in the obstacle detection reliability evaluation method according to the present invention, the storage unit further stores a second lookup table of obstacle pair scores, the second lookup table includes a plurality of obstacle positions related to positions of obstacles in the image, and a plurality of scores respectively corresponding to the obstacle positions:
in step (a), the first current obstacle detection result and the second current obstacle detection result both include the detected obstacle, and the detected obstacle is located at the position of the current image; and
in step (G), the processing unit obtains the scores corresponding to the different obstacles according to the positions of the different obstacles in the current image or the previous image and the second lookup table, and obtains the fourth penalty score according to the obtained scores.
Preferably, in the method for evaluating the confidence of obstacle detection according to the present invention, in step (a), the first classification method detects obstacles by contour texture, and the second classification method detects obstacles by deep learning.
Preferably, in the method for assessing confidence level of obstacle detection according to the present invention, in step (a), the first classification method obtains the contour texture of the current image in a gradient direction histogram and a logarithmic weighting mode, and detects obstacles by using a support vector machine, and the second classification method detects obstacles by using a deep learning convolutional neural network.
The invention has the beneficial effects that: and judging whether at least one different obstacle exists in the first current obstacle detection result and the second current obstacle detection result by the processing unit so as to judge whether at least one penalty score is obtained, and subtracting the at least one penalty score from the confidence score after the at least one penalty score is obtained so as to update the confidence score.
Drawings
Other features and effects of the present invention will become apparent from the following detailed description of the embodiments with reference to the accompanying drawings, in which:
FIG. 1 is a block diagram illustrating a system for implementing the obstacle detection confidence evaluation method of the present invention;
FIG. 2 is a flow chart illustrating the method for assessing confidence in obstacle detection according to this embodiment of the present invention;
fig. 3 is a diagram illustrating a first current obstacle detection result;
fig. 4 is a diagram illustrating a second current obstacle detection result;
FIG. 5 is a flowchart illustrating the sub-steps of step 23 of the embodiment in conjunction with FIG. 2;
FIG. 6 is a schematic diagram illustrating a previous image; and
fig. 7 is a schematic diagram illustrating a first previous obstacle detection result in conjunction with fig. 6;
fig. 8 is a schematic diagram illustrating a second previous obstacle detection result in conjunction with fig. 6;
FIG. 9 is a diagram illustrating a current image;
fig. 10 is a schematic view illustrating another first current obstacle detection result in conjunction with fig. 9; and
fig. 11 is a schematic diagram illustrating another second current obstacle detection result in conjunction with fig. 9.
Detailed Description
Referring to fig. 1, a system 100 for implementing an embodiment of the obstacle detection confidence evaluation method of the present invention is illustrated. The system 100 includes a storage unit 11, an image capturing unit 12, and a processing unit 13 electrically connected to the storage unit 11 and the image capturing unit 12. The image capturing unit 12 continuously captures and transmits a series of images to the processing unit 13. The storage unit 11 stores a confidence score such as 100, a first lookup table of barrier pair scores, a second lookup table of barrier pair scores, a first penalty weight corresponding to a first classification method, and a second penalty weight corresponding to a second classification method. The first lookup table includes a plurality of obstacle distances related to the distance between the actual obstacle and the image capturing unit 12, a plurality of obstacle types related to the types of obstacles, and a plurality of scores respectively corresponding to the obstacle distances and the types of obstacles, and table 1 illustrates the first lookup table. The second lookup table includes a plurality of obstacle positions related to positions of obstacles in an image and a plurality of scores respectively corresponding to the obstacle positions, and table 2 illustrates the second lookup table.
TABLE 1
TABLE 2
Referring to fig. 1 and 2, how the system 100 performs the obstacle detection confidence evaluation method according to the present invention is described. The steps involved in this example are described in detail below.
In step 21, after the processing unit 13 receives a current image from the image capturing unit 12, the processing unit 13 obtains and stores a first current obstacle detection result indicating an obstacle detected by the first classification method according to the current image by using the first classification method (see fig. 3), and obtains and stores a second current obstacle detection result indicating an obstacle detected by the second classification method according to the current image by using the second classification method (see fig. 4). It should be noted that, in the present embodiment, the first classification method obtains the contour texture of the current image by using Histogram of Oriented Gradients (HOG) and Log Weighted Patterns (LWP), and detects the obstacle by using Support Vector Machine (SVM), and the second classification method detects the obstacle by using Convolutional Neural Networks (CNN) of deep learning (deep learning), but not limited thereto. Both the first classification method and the second classification method detect obstacles, not only the obstacles, but also the positions of the obstacles in the current image and mark the types of the detected obstacles (see fig. 3 and 4), so that the first current obstacle detection result and the second current obstacle detection result both include the detected obstacles, the positions of the detected obstacles in the current image, and the types of the detected obstacles.
In step 22, the processing unit 13 determines whether at least one different obstacle exists in the first current obstacle detection result and the second current obstacle detection result. When the processing unit 13 determines that the first current obstacle detection result and the second current obstacle detection result have the at least one different obstacle, performing step 23; and when the processing unit 13 determines that the first current obstacle detection result and the second current obstacle detection result do not have the at least one different obstacle, step 24 is performed.
In step 23, the processing unit 13 obtains at least one penalty score corresponding to each of the at least one different obstacle according to the at least one different obstacle, and subtracts the at least one penalty score from the confidence score to update the confidence score stored in the storage unit 11. In the present embodiment, assuming that there are N different obstacles, N ≧ 1, step 23 includes sub-steps 231-236 (see fig. 5), but not limited thereto.
Referring again to FIGS. 1 and 5, substeps 231 through 236 are further illustrated.
In sub-step 231, initially, the processing unit 13 determines whether the first current obstacle detection result has a 1 st different obstacle for the 1 st different obstacle, that is, i is 1.
In sub-step 232, the processing unit 13 determines whether the first current obstacle detection result has an ith different obstacle. When the processing unit 13 determines that the first current obstacle detection result does not have the ith different obstacle, step 233 is performed; when the processing unit 13 determines that the ith different obstacle exists in the first current obstacle detection result, step 234 is performed.
In sub-step 233, the processing unit 13 obtains a first penalty score corresponding to the ith different obstacle as one of the at least one penalty score according to the ith different obstacle, and subtracts the first penalty score from the confidence score to update the confidence score stored in the storage unit 11. In this embodiment, the processing unit 13 first estimates the distance between the image capturing unit 12 and the actual obstacle corresponding to the ith different obstacle according to the position of the ith different obstacle in the current image by using a known image distance estimation method, then obtains a score corresponding to the ith different obstacle by matching with the first lookup table according to the estimated distance between the image capturing unit 12 and the actual obstacle corresponding to the ith different obstacle and the type of the ith different obstacle, and then multiplies the obtained score by the first penalty weight to obtain the first penalty score; however, in other embodiments, the processing unit 13 may also obtain the score corresponding to the ith different obstacle according to other attributes of the ith different obstacle, such as obstacle position, etc., in conjunction with another lookup table related to other obstacle attributes and scores, and then multiply the obtained score by the first penalty weight to obtain the first penalty score; even the processing unit 13 may set the i-th different obstacle as the default score without matching the first lookup table, and then multiply the default score by the first penalty weight to obtain the first penalty score, which is not limited to this. It is worth mentioning that since the features of the present invention are not known to those skilled in the art as the image distance estimation method, their details are omitted here for the sake of brevity.
In sub-step 234, the processing unit 13 obtains a second penalty score corresponding to the ith different obstacle as one of the at least one penalty score according to the ith different obstacle, and subtracts the second penalty score from the confidence score to update the confidence score stored in the storage unit 11. In this embodiment, the processing unit 13 first estimates the distance between the actual obstacle corresponding to the ith different obstacle and the image capturing unit 12 according to the position of the ith different obstacle in the current image by using the image distance estimation method, then obtains a score corresponding to the ith different obstacle by using the estimated distance between the actual obstacle corresponding to the ith different obstacle and the image capturing unit 12 and the type of the ith different obstacle, and obtains the second penalty score by matching with the first lookup table and multiplying the obtained score by the second penalty weight; however, in other embodiments, the processing unit 13 may also obtain the score corresponding to the ith different obstacle according to other attributes of the ith different obstacle, such as obstacle position, etc. in cooperation with another lookup table related to other obstacle attributes and scores, and then multiply the obtained score by the second penalty weight to obtain the second penalty score; even the processing unit 13 may set the i-th different obstacle as the default score without matching the first lookup table, and then multiply the default score by the second penalty weight to obtain the second penalty score, which is not limited to this.
In sub-step 235 following sub-steps 233 and 234, the processing unit 13 determines whether the i-th different obstacle is an nth different obstacle, that is, whether i is equal to N. When the processing unit 13 determines that the ith different obstacle is the nth different obstacle, performing step 24; and when the processing unit 13 determines that the ith different obstacle is not the nth different obstacle, performing substep 236.
In sub-step 236, the processing unit 13 sets i to i +1 for the (i + 1) th different obstacle. Substeps 232-236 are then repeated until i ═ N.
In step 24, the processing unit 13 determines whether there is at least one different obstacle in the first previous obstacle detection result and the first current obstacle detection result according to a first previous obstacle detection result indicating the obstacle detected by the first classification method and the first current obstacle detection result obtained by the processing unit 13 according to a previous image, and performs step 25 when the processing unit 13 determines that there is the at least one different obstacle in the first previous obstacle detection result and the first current obstacle detection result; when the processing unit 13 determines that the first previous obstacle detection result and the first current obstacle detection result do not have the at least one different obstacle, step 26 is performed.
In step 25, the processing unit 13 obtains at least one third penalty score corresponding to the at least one different obstacle according to the at least one different obstacle, the second lookup table and the first penalty weight, and subtracts the at least one third penalty score from the confidence score to update the confidence score stored in the storage unit 11. In this embodiment, for each different obstacle, the processing unit 13 matches the second lookup table according to the position of the different obstacle in the current image or the previous image to obtain a score corresponding to the different obstacle, and then multiplies the obtained score by the first penalty weight to obtain the third penalty score; however, in other embodiments, the processing unit 13 may also obtain the score corresponding to the different obstacle according to other attributes of the different obstacle, such as the obstacle type or the obstacle distance, and obtain the third penalty score by multiplying the obtained score by the first penalty weight; even the processing unit 13 may set each of the different obstacles as the same default score without matching the second lookup table, and then multiply the default score by the first penalty weight to obtain the third penalty score, which is not limited thereto. For example, if the first previous obstacle detection result indicates an obstacle a and an obstacle b, and the first current obstacle detection result indicates the obstacle a and an obstacle c, the processing unit 13 determines that two different obstacles of the obstacle b and the obstacle c exist in the first previous obstacle detection result and the first current obstacle detection result.
In step 26, the processing unit 13 determines whether there is at least one different obstacle in the second previous obstacle detection result and the second current obstacle detection result according to a second previous obstacle detection result indicating the obstacle detected by the second classification method and the second current obstacle detection result obtained by the processing unit 13 according to the previous image, and performs step 27 when the processing unit 13 determines that there is the at least one different obstacle in the second previous obstacle detection result and the second current obstacle detection result; when the processing unit 13 determines that the second previous obstacle detection result and the second current obstacle detection result do not have the at least one different obstacle, step 28 is performed.
In step 27, the processing unit 13 obtains at least one fourth penalty score corresponding to the at least one different obstacle according to the at least one different obstacle, the second lookup table and the second penalty weight, and subtracts the at least one fourth penalty score from the confidence score to update the confidence score stored in the storage unit 11. In this embodiment, for each different obstacle, the processing unit 13 matches the second lookup table according to the position of the different obstacle in the current image or the previous image to obtain a score corresponding to the different obstacle, and then multiplies the obtained score by the second penalty weight to obtain the fourth penalty score; however, in other embodiments, the processing unit 13 may also obtain the score corresponding to the different obstacle according to other attributes of the different obstacle, such as the obstacle type or the obstacle distance, and obtain the fourth penalty score by multiplying the obtained score by the second penalty weight; even the processing unit 13 may set each of the different obstacles as the same default score without matching the second lookup table, and then multiply the default score by the second penalty weight to obtain the fourth penalty score, which is not limited thereto. It is noted that in the present embodiment, steps 22 and 23 are before steps 24 to 27, and in other embodiments, steps 24 to 27 may be before steps 22 and 23, that is, step 22 is executed when the determination result of step 26 is negative, and step 28 is executed when the determination result of step 22 is negative.
In step 28, the processing unit 13 outputs the confidence score and then resets the confidence score to 100. It should be noted that in the present embodiment, if the confidence score is a negative number, the processing unit 13 corrects the confidence score to 0 and outputs the corrected confidence score, but the present invention is not limited thereto, and the processing unit 13 may also directly output the confidence score with a negative sign.
Since the third penalty score and the fourth penalty score are related to different obstacles in the first previous obstacle detection result and the first current obstacle detection result and different obstacles in the second previous obstacle detection result and the second current obstacle detection result, the third penalty score and the fourth penalty score are related to different obstacles in the previous image or the current image, so as to determine whether the different obstacles enter or exit the shooting range of the image shooting unit 12 at different times or whether the detection of the system 100 has a serious error. And since the first classification method corresponds to the first penalty score and the third penalty score, and the second classification method corresponds to the second penalty score and the fourth penalty score, the first penalty score and the third penalty score correspond to a first penalty weight, the second penalty score and the fourth penalty score correspond to a second penalty weight, the first penalty weight is 0.2 for example, and the second penalty weight is 0.8 for example.
Referring to fig. 1, 6 and 7, an embodiment of the obstacle detection reliability evaluation method of the present invention will be described below with reference to an application example, fig. 6 illustrates the previous image including an obstacle a, an obstacle B, and an obstacle C, fig. 7 illustrates a first previous obstacle detection result obtained by the processing unit 13 according to the previous image by using the first classification method, the first previous obstacle detection result indicates that the obstacle a is located in the middle of the previous image and the current image, the type is vehicle, the distance is 12 meters from the image capturing unit, and the obstacle B is located at the edge of the previous image, the type is vehicle, the distance is 12 thirty meters from the image capturing unit, fig. 8 illustrates a second previous obstacle detection result obtained by the processing unit 13 according to the previous image by using the second classification method, the second previous obstacle detection result indicates that the obstacle a is located in the previous image and the current image The type of the obstacle is a car, which is 12 meters away from the image capturing unit, the obstacle B is located at the edge of the previous image, the type of the obstacle is a car, which is 12 thirty meters away from the image capturing unit, and the obstacle C is located at the middle partial edge of the previous image and the current image, the type of the obstacle is a person, which is 12 thirty meters away from the image capturing unit. Fig. 9 illustrates the current image including the obstacle a and the obstacle C, fig. 10 illustrates a first current obstacle detection result obtained by the processing unit 13 according to the current image by the first classification method, the first current obstacle detection result indicating that the obstacle a is located in the middle of the current image and the current image, the category being a vehicle, which is 12 meters away from the image capturing unit, fig. 11 illustrates a second current obstacle detection result obtained by the processing unit 13 according to the current image by the second classification method, the second current obstacle detection result indicating that the obstacle a is located in the middle of the current image, the category being a vehicle, which is 12 meters away from the image capturing unit, and the obstacle C is located at a middle partial edge of the current image, the category being a person, which is thirty meters away from the image capturing unit 12.
As shown in step 233, the processing unit 13 obtains a score corresponding to the obstacle C (the score is 70 because the obstacle is human and the obstacle distance is thirty meters) according to the obstacle type and the obstacle distance of the different obstacle (the obstacle C) indicated by the first current obstacle detection result and the first lookup table, and multiplies the obtained score (i.e., 70) by the first penalty weight (i.e., 0.2) to obtain the first penalty score (i.e., 0.2 × 70 ═ 14), and the processing unit 13 updates the confidence score to be 100-14 ═ 86. As shown in step 25, the processing unit 13 obtains a score corresponding to the obstacle B (the score is 0 because the obstacle B is located at an edge) according to the position of the different obstacle (the obstacle B) in the previous image and the second lookup table, and multiplies the obtained score (that is, 0) by the first penalty weight to obtain the third penalty score (that is, 0.2 × 0 equals to 0), and the processing unit 13 updates the confidence score to 86-0 equals to 86. As shown in step 27, the processing unit 13 obtains a score corresponding to the obstacle B (the score is 0 because the obstacle B is located at an edge) according to the position of the different obstacle (the obstacle B) in the previous image and the second lookup table, multiplies the obtained score (that is, 0) by the second penalty weight (that is, 0.8) to obtain the fourth penalty score (that is, 0.8 × 0 ═ 0), and the processing unit 13 updates the confidence score to 86-0 ═ 86. Finally, in step 28, the processing unit 13 outputs the confidence score as 86.
To sum up, the method for evaluating the confidence level of obstacle detection according to the present invention determines whether there is at least one obstacle different from the first previous obstacle detection result and the first current obstacle detection result, whether there is at least one obstacle different from the second previous obstacle detection result and the second current obstacle detection result, and whether there is at least one obstacle different from the first current obstacle detection result and the second current obstacle detection result, so as to determine whether to obtain at least one first penalty score, at least one second penalty score, at least one third penalty score, and at least one fourth penalty score, and after obtaining the penalty scores, subtracts the penalty scores from the confidence level to update and output the confidence level score, thereby providing the confidence level of the obstacle detection result for the driving reference, therefore, the object of the present invention can be achieved.
The above description is only an example of the present invention, and the scope of the present invention should not be limited thereby, and the invention is still within the scope of the present invention by simple equivalent changes and modifications made according to the claims and the contents of the specification.
Claims (10)
1. An obstacle detection reliability evaluation method implemented by a processing unit electrically connected to a storage unit and an image capturing unit, the storage unit storing a reliability score, the image capturing unit continuously capturing and transmitting a series of images to the processing unit, the method comprising: the obstacle detection reliability evaluation method comprises the following steps:
(A) after receiving a current image from the image capturing unit, obtaining and storing a first current obstacle detection result indicating an obstacle detected by the first classification method by using a first classification method according to the current image, and obtaining and storing a second current obstacle detection result indicating an obstacle detected by the second classification method by using a second classification method according to the current image;
(B) determining whether at least one different obstacle exists in the first current obstacle detection result and the second current obstacle detection result; and
(C) when the first current obstacle detection result and the second current obstacle detection result are judged to have the at least one different obstacle, at least one penalty score corresponding to the at least one different obstacle is obtained according to the at least one different obstacle, and the at least one penalty score is subtracted from the confidence score to update the confidence score.
2. The obstacle detection reliability evaluation method according to claim 1, wherein: step (C) comprises the sub-steps of:
(C-1) when it is determined that the first current obstacle detection result and the second current obstacle detection result have the at least one different obstacle, determining, for each of the at least one different obstacle, whether the first current obstacle detection result does not have the different obstacle;
(C-2) when it is determined that the different obstacle does not exist in the first current obstacle detection result, obtaining a first penalty score, which is one of the at least one penalty score, according to at least the different obstacle, and subtracting the first penalty score from the confidence score to update the confidence score; and
(C-3) when it is determined that the different obstacle exists in the first current obstacle detection result, obtaining a second penalty score, which is one of the at least one penalty score, according to at least the different obstacle, and subtracting the second penalty score from the confidence score to update the confidence score.
3. The obstacle detection reliability evaluation method according to claim 2, wherein: the storage unit further stores a first lookup table of obstacle pair scores, a first penalty weight corresponding to the first classification method, and a second penalty weight corresponding to the second classification method, the first lookup table including a plurality of obstacle distances related to distances of actual obstacles from the image capturing unit, and a plurality of scores respectively corresponding to the obstacle distances, wherein:
in step (a), the first current obstacle detection result and the second current obstacle detection result both include the detected obstacle, and the detected obstacle is located at the position of the current image;
in step (C-2), the processing unit estimates a distance between an actual obstacle corresponding to the different obstacle and the image capturing unit according to the position of the different obstacle in the current image, obtains a score corresponding to the different obstacle according to the estimated distance and the first lookup table, and multiplies the obtained score by the first penalty weight to obtain the first penalty score; and
in step (C-3), the processing unit estimates a distance between an actual obstacle corresponding to the different obstacle and the image capturing unit according to the position of the different obstacle in the current image, obtains a score corresponding to the different obstacle according to the estimated distance and the first lookup table, and multiplies the obtained score by the second penalty weight to obtain the second penalty score.
4. The obstacle detection reliability evaluation method according to claim 2, wherein: the storage unit further stores a first lookup table of obstacle pair scores, a first penalty weight corresponding to the first classification method, and a second penalty weight corresponding to the second classification method, the first lookup table including a plurality of obstacle categories and a plurality of scores respectively corresponding to the categories, wherein:
in the step (a), the first current obstacle detection result and the second current obstacle detection result both include the detected obstacle and the type of the detected obstacle;
in step (C-2), the processing unit obtains the scores corresponding to the different obstacles according to the types of the different obstacles and the first lookup table, and multiplies the obtained scores by the first penalty weight to obtain the first penalty score; and in the step (C-3), the processing unit obtains the scores corresponding to the different obstacles according to the types of the different obstacles and the first lookup table, and multiplies the obtained scores by the second penalty weight to obtain the second penalty score.
5. The obstacle detection reliability evaluation method according to claim 1, wherein: after the step (A), further comprising the steps of:
(D) determining whether at least one different obstacle exists in a first previous obstacle detection result and a first current obstacle detection result, which are obtained according to a previous image and indicate the obstacle detected by the first classification method;
(E) when the first previous obstacle detection result and the first current obstacle detection result are judged to have the at least one different obstacle, at least one third penalty score corresponding to the at least one different obstacle is obtained according to the at least one different obstacle, and the confidence score is subtracted by the at least one third penalty score so as to update the confidence score.
6. The obstacle detection reliability evaluation method according to claim 5, wherein: the storage unit further stores a second lookup table of obstacle pair scores, the second lookup table including a plurality of obstacle positions related to positions of obstacles in the image, and a plurality of scores respectively corresponding to the obstacle positions, wherein:
in step (a), the first current obstacle detection result and the second current obstacle detection result both include the detected obstacle, and the detected obstacle is located at the position of the current image; and
in step (E), the processing unit obtains the scores corresponding to the different obstacles according to the positions of the different obstacles in the current image or the previous image and the second lookup table, and obtains the third penalty score according to the obtained scores.
7. The obstacle detection reliability evaluation method according to claim 1, wherein: after step (a), further comprising the steps of:
(F) determining whether at least one different obstacle exists in a second previous obstacle detection result and a second current obstacle detection result, which are obtained according to a previous image and indicate the obstacle detected by the second classification method;
(G) when the second previous obstacle detection result and the second current obstacle detection result are judged to have the at least one different obstacle, at least one fourth penalty score corresponding to the at least one different obstacle is obtained according to the at least one different obstacle, and the confidence score is subtracted by the at least one fourth penalty score so as to update the confidence score.
8. The obstacle detection reliability evaluation method according to claim 7, wherein: the storage unit further stores a second lookup table of obstacle pair scores, the second lookup table including a plurality of obstacle positions related to positions of obstacles in the image, and a plurality of scores respectively corresponding to the obstacle positions, wherein:
in step (a), the first current obstacle detection result and the second current obstacle detection result both include the detected obstacle, and the detected obstacle is located at the position of the current image; and
in step (G), the processing unit obtains the scores corresponding to the different obstacles according to the positions of the different obstacles in the current image or the previous image and the second lookup table, and obtains the fourth penalty score according to the obtained scores.
9. The obstacle detection reliability evaluation method according to claim 1, wherein: in the step (a), the first classification method detects the obstacle by contour texture, and the second classification method detects the obstacle by deep learning.
10. The obstacle detection reliability evaluation method according to claim 9, wherein: in the step (a), the first classification method obtains the current image contour texture by using a gradient direction histogram and a logarithmic weighting mode, and detects the obstacle by using a support vector machine, and the second classification method detects the obstacle by using a deep learning convolutional neural network.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711431445.5A CN109960990B (en) | 2017-12-26 | 2017-12-26 | Method for evaluating reliability of obstacle detection |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201711431445.5A CN109960990B (en) | 2017-12-26 | 2017-12-26 | Method for evaluating reliability of obstacle detection |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109960990A CN109960990A (en) | 2019-07-02 |
| CN109960990B true CN109960990B (en) | 2022-06-03 |
Family
ID=67021870
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201711431445.5A Active CN109960990B (en) | 2017-12-26 | 2017-12-26 | Method for evaluating reliability of obstacle detection |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109960990B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110568851A (en) * | 2019-09-30 | 2019-12-13 | 重庆元韩汽车技术设计研究院有限公司 | Automobile chassis motion control system and method based on remote control |
| DE102020101482B3 (en) * | 2020-01-22 | 2021-04-15 | Sick Ag | Security system and procedures |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2005121496A (en) * | 2003-10-17 | 2005-05-12 | Toyota Motor Corp | Obstruction detecting radar installation of vehicle |
| CN103043000A (en) * | 2011-10-13 | 2013-04-17 | 能晶科技股份有限公司 | Obstacle detection system and obstacle detection method thereof |
| CN105095848A (en) * | 2014-12-01 | 2015-11-25 | 北京云识图信息技术有限公司 | Object identification method and system |
| CN105372654A (en) * | 2015-12-14 | 2016-03-02 | 财团法人车辆研究测试中心 | Obstacle classification reliability quantification method |
| CN107240299A (en) * | 2017-07-20 | 2017-10-10 | 北京纵目安驰智能科技有限公司 | Autonomous land vehicle is to mobile object identification and the method for vehicle obstacle-avoidance |
-
2017
- 2017-12-26 CN CN201711431445.5A patent/CN109960990B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2005121496A (en) * | 2003-10-17 | 2005-05-12 | Toyota Motor Corp | Obstruction detecting radar installation of vehicle |
| CN103043000A (en) * | 2011-10-13 | 2013-04-17 | 能晶科技股份有限公司 | Obstacle detection system and obstacle detection method thereof |
| CN105095848A (en) * | 2014-12-01 | 2015-11-25 | 北京云识图信息技术有限公司 | Object identification method and system |
| CN105372654A (en) * | 2015-12-14 | 2016-03-02 | 财团法人车辆研究测试中心 | Obstacle classification reliability quantification method |
| CN107240299A (en) * | 2017-07-20 | 2017-10-10 | 北京纵目安驰智能科技有限公司 | Autonomous land vehicle is to mobile object identification and the method for vehicle obstacle-avoidance |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109960990A (en) | 2019-07-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20200272835A1 (en) | Intelligent driving control method, electronic device, and medium | |
| CN106960446B (en) | Unmanned ship application-oriented water surface target detection and tracking integrated method | |
| US20220245952A1 (en) | Parking spot detection method and parking spot detection system | |
| US11170525B2 (en) | Autonomous vehicle based position detection method and apparatus, device and medium | |
| US9298995B2 (en) | Red-eye determination device | |
| CN110390306A (en) | Detection method, vehicle and the computer readable storage medium of right angle parking stall | |
| CN111553302B (en) | Key frame selection method, device, equipment and computer readable storage medium | |
| CN111126209B (en) | Lane line detection method and related equipment | |
| CN111967396A (en) | Processing method, device and equipment for obstacle detection and storage medium | |
| CN109960990B (en) | Method for evaluating reliability of obstacle detection | |
| JP2021033510A (en) | Driving assistance device | |
| CN115797736A (en) | Method, device, equipment and medium for training target detection model and target detection | |
| CN112101303A (en) | Image data processing method and device and computer readable storage medium | |
| US10467486B2 (en) | Method for evaluating credibility of obstacle detection | |
| CN113420704A (en) | Object identification method and device based on visual sensor and robot | |
| TWI645997B (en) | Obstacle detection credibility evaluation method | |
| CN111402610A (en) | Method, device, equipment and storage medium for identifying lighting state of traffic light | |
| CN116343143A (en) | Target detection method, storage medium, road side equipment and automatic driving system | |
| CN112818845B (en) | Test method, target object detection method, driving control method and driving control device | |
| CN113326715B (en) | Target association method and device | |
| CN111985267B (en) | Multi-target detection method, device, computer equipment, vehicle and storage medium | |
| CN112686155A (en) | Image recognition method, image recognition device, computer-readable storage medium and processor | |
| CN111583387A (en) | Method and system for three-dimensional reconstruction of outdoor scene of unmanned aerial vehicle | |
| CN117152707B (en) | Calculation method and device for offset distance of vehicle and processing equipment | |
| CN111260698A (en) | Binocular image feature matching method and vehicle-mounted terminal |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |