CN113228130B - image processing device - Google Patents

Abstract

The present invention realizes an image processing device capable of accurately identifying whether a pedestrian is a pedestrian even when the whole body of a pedestrian existing in the surrounding area cannot be detected. In the camera image (1101) in which the feet are cut off during the daytime, the result of correcting the recognition region (11020) so that the upper body including the pedestrian's head is the recognition region becomes the recognition region (1102). For the camera image (1103) in which the headlight of the own vehicle is low beam at night, it can be judged that the head of the pedestrian is not included in the detection area (1103), in order to use the upper body including the head of the pedestrian as the recognition area, The recognition region (11030) is corrected as the recognition region (1104). In the nighttime camera image (1105) in which the headlights of the own vehicle are on high beam, the image (11060) in which the foot of the pedestrian is cut off is corrected so that the upper body including the head of the pedestrian is the recognition region, and becomes the recognition region (1106).

Description

image processing device

technical field

The invention relates to an image processing device for identifying pedestrians in a vehicle-mounted camera.

Background technique

In recent years, the development of driver assistance systems and automated driving systems has been progressing due to the spread of in-vehicle sensor devices. In more advanced system development, it is necessary to recognize whether objects existing around the car are vehicles or pedestrians.

As described in Patent Document 1, a method of recognizing objects detected by an on-vehicle camera, radar, etc. using a classifier generated in advance by machine learning is used to recognize the type of objects existing around a car.

prior art literature

patent documents

Patent Document 1: Japanese Patent Laid-Open No. 2013-232080

Contents of the invention

The problem to be solved by the invention

Pedestrian recognition in the prior art is based on the premise that the whole body of the pedestrian is reflected in the detected area, and recognition performance is maintained by performing recognition processing of the whole body of the pedestrian.

Therefore, when there are pedestrians at close range such that the lower body of the pedestrian is exposed outside the viewing angle, the lower body of the pedestrian is blocked by hedges, guardrails, etc., and the whole body of the pedestrian cannot be detected due to the illumination conditions of the headlights at night, etc., the pedestrian Recognition performance will degrade.

An object of the present invention is to realize an image processing device capable of accurately identifying whether a pedestrian is a pedestrian even when the whole body of a pedestrian existing in the surrounding area cannot be detected.

technical means to solve problems

In order to achieve the above objects, the present invention is constituted as follows.

The image processing device includes: a detection unit that detects image information of an external environment; a recognition area correction amount calculation unit that calculates a correction amount for a recognition area that recognizes a three-dimensional object based on the image information detected by the detection unit; a recognition area correction section that corrects the recognition area of the image information for recognizing the three-dimensional object based on the correction amount calculated by the recognition area correction amount calculation section; The recognition process is performed on the three-dimensional objects within the recognition region corrected by the recognition region correction unit.

The effect of the invention

According to the present invention, it is possible to realize an image processing device capable of accurately identifying whether a pedestrian is a pedestrian even when the whole body of a pedestrian existing in the surrounding area cannot be detected.

Description of drawings

FIG. 1 is a block diagram showing the overall configuration of an image processing device according to an embodiment of the present invention.

FIG. 2 is a diagram showing a processing flow in the stereo camera device.

FIG. 3 is a diagram illustrating the result of a three-dimensional object detection process on a camera image.

FIG. 4 is a diagram showing a processing configuration of an embodiment.

FIG. 5 is a diagram showing details of a processing configuration of an embodiment.

FIG. 6 is a diagram showing an example of nighttime camera images and operation results of three-dimensional object detection processing.

FIG. 7 is a diagram showing a three-dimensional object detection result when an obstructer is included in the detection area in the three-dimensional object detection process.

FIG. 8 is a diagram showing an image in a state where the foot of a three-dimensional object is cut off.

FIG. 9 is a diagram showing the relationship between the angle of view of a camera and the distance of pedestrians.

FIG. 10 is a diagram showing the relationship between the headlight irradiation area and the distance of pedestrians.

FIG. 11 is a diagram showing an example of a recognition region correction processing result.

detailed description

Embodiments of the present invention will be described using the drawings.

Example

FIG. 1 is a block diagram showing the overall configuration of an image processing device according to an embodiment of the present invention.

In FIG. 1 , an in-vehicle stereo camera device 100 according to an embodiment of the present invention is mounted on a vehicle and recognizes the environment outside the vehicle based on image information of the external environment of the imaging target area in front of the vehicle. The vehicle-mounted stereo camera device 100, for example, recognizes white lines on the road, pedestrians, vehicles, other three-dimensional objects, signals, signs, and lighting lamps, and performs braking and steering adjustments of the vehicle (own vehicle) on which the stereo camera device 100 is mounted. and other adjustments.

The vehicle-mounted stereo camera device 100 is provided with: two cameras (a left camera 101 and a right camera 102 (an imaging section that captures an image of the external environment) arranged on the left and right to acquire image information, and are used to control the shooting of the cameras 101 and 102, and capture The image input interface 103 for captured images.

Images taken in through the image input interface 103 are sent data via the bus 109 , processed by the image processing unit 104 and the arithmetic processing unit 105 , and image data and the like that are in-process and final results are stored in the storage unit 106 .

The image processing unit 104 compares the first image obtained from the imaging element of the camera 101 with the second image obtained from the imaging element of the camera 102, and performs device-specific processing by the imaging elements of the camera 101 and the camera 102 on each image. Image corrections such as offset correction and noise interpolation are stored in the storage unit 106 .

In addition, the positions corresponding to each other between the first image from the camera 101 and the second image from the camera 102 are calculated to calculate parallax information, and the parallax information is stored in the storage unit 106 in the same manner as above.

The arithmetic processing unit 105 uses the image and parallax information (distance information with respect to each point on the image) stored in the storage unit 106 to perform recognition of various objects necessary for sensing the surrounding environment of the vehicle.

The various objects refer to people, cars, other obstacles, signal lights, signs, rear lights and headlights of cars, and the like. These identification results and part of intermediate calculation results are recorded in the storage unit 106 in the same manner as described above. After performing various object recognitions on captured images, these recognition results are used to calculate command values and the like required for vehicle control.

The control policy of the vehicle obtained as a calculation result and part of the object recognition result are transmitted to the in-vehicle network CAN 110 through the CAN interface 107, thereby performing braking of the vehicle.

Furthermore, the control processing unit 108 is configured to prevent abnormal operations by monitoring whether abnormal operations have occurred in each processing unit or whether an error has occurred during data transmission, etc. regarding these operations.

The above-mentioned image processing unit 104 is connected to the control processing unit 108, the storage unit 106, the arithmetic processing unit 105, the input and output unit 103 between the imaging elements of the left camera 101 and the right camera 102, and the input to the external vehicle network CAN110 via the internal bus 109. Output section 107.

The image input interface 103 , the image processing unit 104 , the storage unit 106 , the arithmetic processing unit 105 , the input and output unit 107 , and the control processing unit 108 are composed of a single or a plurality of computer units.

The storage unit 106 is constituted by a memory or the like which stores, for example, image information obtained by the image processing unit 104 , image information created based on a result of scanning by the arithmetic processing unit 105 , and the like.

The input/output unit 107 with the external vehicle network CAN 110 outputs the information output from the vehicle stereo camera device 100 to the control system (not shown) of the own vehicle via the external vehicle network CAN 110 .

FIG. 2 is a diagram showing the flow of processing in the stereo camera device 100 .

First, in the vehicle-mounted stereo camera device 100, images are captured by the left and right cameras 101 and 102, and the image processing unit 104 performs image processing 205 such as correction for absorbing the inherent characteristics of the imaging device on the image data 203 and 204 captured respectively. The processing result is stored in the image buffer 206 . The image buffer 206 is provided in the storage unit 106 in FIG. 1 .

In addition, the parallax information of the images obtained by the left and right cameras is obtained by comparing the images using the corrected two images. From the parallax of the left and right images, it becomes clear where and where a point of interest on the target object corresponds to on the images of the left and right cameras 101 and 102 , and the distance to the target object can be obtained based on the principle of triangulation.

What performs this kind of processing is the parallax processing 207 . Image processing 205 and parallax processing 207 are performed by the image processing unit 104 in FIG. 1 , and the finally obtained images and parallax information are stored in the storage unit 106 .

Using the parallax images obtained as described above, detection processing 208 for detecting a three-dimensional object in a three-dimensional space is performed. Furthermore, various object recognition processing 209 is performed using the above-described images and parallax information stored in the storage unit 106 . Objects to be recognized include people, cars, other three-dimensional objects, signs, signal lights, and rear lights. The details of the recognition process are determined by constraints such as the characteristics of the object and the processing time spent on the system.

Furthermore, in consideration of the result of the object recognition and the state of the host vehicle (speed, steering angle, etc.), it is decided that the vehicle control process 210, for example, issue a warning to the occupant, and perform controls such as braking and steering angle adjustment of the host vehicle. A guideline for moving, or a guideline for controlling object avoidance based on this, and the result is output through the CAN interface 108.

The three-dimensional object detection processing 208 , the various object recognition processing 209 , and the vehicle control processing 210 are performed by the arithmetic processing unit 105 in FIG. 1 . Each of these processes and methods is composed of, for example, a single or a plurality of computer units, and is configured to be able to exchange data with each other.

The parallax or distance of each pixel of the left and right images is obtained by the above-mentioned parallax processing 207, which is combined into a three-dimensional object in the three-dimensional object detection processing unit 208, and various object recognition processing 209 is performed according to the position and area on the image. At this time, in order for the various object recognition processing 209 to recognize objects stably, it is necessary to match the three-dimensional object region on the image with the image of the object to be recognized.

However, in a stereo camera, it may not be possible to completely match the object area on the image to be recognized due to the brightness of the external environment, variation in imaging performance between cameras, occlusion by foreign matter on the glass surface, or the like.

This is the same even when a radar such as a millimeter wave is combined with an image sensor such as a camera. Therefore, in various identification processes, one embodiment of the present invention is applied.

The configuration will be described below on the premise of a stereo camera device.

FIG. 3 is a diagram illustrating the results of the stereoscopic object detection process 208 on the camera image.

In FIG. 3 , for each of objects with a height above the road surface such as pedestrians, vehicles, trees, and street lamps that exist in the three-dimensional space, a three-dimensional object region 301 as a result of the three-dimensional object detection process 208 is obtained, which is used as an object on the image Area projection.

The three-dimensional object region 301 may be rectangular as shown in FIG. 3 , or may be an amorphous region obtained according to parallax and distance. In subsequent processing, it is generally handled as a rectangle in order to facilitate computer processing. In the present embodiment, the details of each processing will be described below by treating an area as a rectangle and using a pedestrian as an example of a three-dimensional object.

FIG. 4 is a diagram showing the processing structure of this embodiment. As described above, various object recognition 209 is performed by the arithmetic processing unit 105 . Therefore, the configuration shown in FIG. 4 is the internal configuration of the arithmetic processing unit 105 .

In addition, the configuration will be described below on the premise of a stereo camera.

In FIG. 4 , the detection unit 401 detects the distance from the imaging unit to the three-dimensional object based on the image information of the external environment captured by the cameras 101 and 102 as the imaging unit. The distance distribution and the like of the detection target object region obtained in the detection processing 208 detects whether or not an obstruction or the like is included in the detection target object region.

The area correction amount calculation unit (recognition area correction amount calculation unit) 402 calculates the correction amount of the recognition area for recognizing a three-dimensional object based on the image information detected by the detection unit 401 . That is, the correction amount of the recognition area of the three-dimensional object is calculated according to at least one of the detected brightness of the external environment, the distance to the three-dimensional object, the illumination range of the headlight, and the position or size of the occluder.

In the recognition area correction unit 403 , correction of the recognition area is performed based on the correction amount calculated by the area correction amount calculation unit 402 .

In the recognition processing unit 404, three-dimensional objects within the set (corrected) recognition area are recognized. Examples of the recognition processing include the following techniques.

There are template matching that compares a pre-prepared template with characteristics of the object to be recognized with the recognition area, using feature quantities such as luminance images or HOG or Haar-Like, and machine learning methods such as support vector machines, Ada-Boost, and Deep Learning Combine the recognizer (software) method.

In addition, an edge shape and the like may be recognized by an artificially determined threshold value judgment.

FIG. 5 is a diagram showing details of the processing configuration of this embodiment (details of each unit shown in FIG. 4 ).

In FIG. 5 , the external environment determination process 501 of the detection unit 401 determines the state of the brightness of the external environment when the three-dimensional object detection process 208 is performed. FIG. 6 is a diagram showing an example of an external environment, and is a diagram showing a nighttime camera image and a camera image 601 as an example of an operation result of the three-dimensional object detection processing unit 208 .

As shown in FIG. 3 , if the external environment is daytime, the entirety of pedestrians detected to move into the image is included in a three-dimensional object area (detection area) 301 . However, when it is dark at night, as shown in FIG. 6, the pedestrian and the background cannot be separated. In the three-dimensional object detection process 208, it is assumed that the detection area 602 includes the lower half of the pedestrian, but does not include the upper half. Entire body that includes pedestrians.

In judging the brightness of the external environment, the brightness distribution in the image, etc., may be obtained by image processing, and the time and the like may be directly obtained from the internal information of the vehicle to make the judgment. In addition, it may be determined that it is daytime if the shutter speed of the camera for which exposure adjustment is performed is below a threshold, and it may be determined as nighttime if it is above the threshold.

In addition, not only the shutter speed can be observed, but also the brightness distribution in the image, the brightness change between frames before and after, etc. can be observed, and detailed external environment judgments such as tunnels and dusk can be carried out.

In FIG. 5 , the obstacle detection process 502 of the detection unit 401 detects whether objects other than pedestrians are included in the detection area when the three-dimensional object detection process 208 is performed, and determines whether the pedestrians are blocked.

FIG. 7 is a diagram showing a three-dimensional object detection result when an obstructer is included in the detection area 701 in the three-dimensional object detection process 208 . In FIG. 7 , observe the parallax distribution 702 of the detection area (recognition area) 701, and when there are many areas with different parallax, calculate the distance of each area, and judge that the pedestrian (three-dimensional object) is blocked by the occluder 703, and then In the detection area, the area of the obstructer 703 and the pedestrian is identified by the area located in the front direction and the depth direction. Then, the correction amount of the recognition area is calculated so as to exclude the obstructing object 703 from the detection area (recognition area) 701, and the three-dimensional object in the corrected recognition area is recognized.

In FIG. 5 , in the distance determination process 503 of the area correction amount calculation unit 402, when the detected three-dimensional object is a pedestrian, it is determined whether it is located at a distance cutting off the foot of the pedestrian, and the correction amount of the recognition area is calculated. . FIG. 8 is a diagram showing an image of a state where the feet of a three-dimensional object (in FIG. 8 , a pedestrian) are cut off. As shown in FIG. 8 , whether to cut off the foot of the three-dimensional object can be determined by whether the lower end of the detection area 801 on the image touches the lower end of the captured image.

In addition, FIG. 9 is a diagram showing the relationship between the angle of view of the camera and the distance between pedestrians. As shown in Figure 9, when the three-dimensional object is a pedestrian, the distance to cut off the feet from the image can be calculated based on the detected distance between the three-dimensional object and the vehicle, the camera shooting angle 901, the height of the camera, and the posture of the camera method etc.

The distance to cut off the foot from the image is preset as a distance setting value and stored in the storage unit 106. As a condition for cutting off the distance to the foot from the image, it is based on the distance detected by the imaging unit (camera 101, camera 102). It is judged whether it is below the distance setting value.

When the detected distance is less than the distance setting value, and it can be judged that the upper part (head) of the detected three-dimensional object is not in the recognition area, the recognition area correcting unit 403 expands (extends) the upper area of the recognition area. Correction. In this case, the recognition region correction unit 403 performs correction so that the correction amount of the upper end region of the recognition region decreases as the detected distance increases within the range where the detected distance is equal to or less than the distance setting value.

In FIG. 5 , the headlight irradiation range determination and correction processing 504 of the area correction amount calculation unit 402 assumes that the pedestrian's head is included in the recognition area based on the illumination state of the vehicle headlight (headlight illuminating the front of the external environment). Calculate the correction amount of the recognition area in the way of . FIG. 10 is a diagram showing the relationship between the headlight irradiation area and the distance of pedestrians. As shown in (A) of FIG. 10 , when the headlight is low beam (when the irradiation range of the headlight is set low), the optical axis of the headlight faces the direction of illuminating the road surface, so pedestrians The closer to the vehicle, the closer the light irradiation range 1001 is to the foot side, and the vicinity of the head is no longer irradiated.

Therefore, as shown in (B) of FIG. 10 , when a pedestrian is observed with the camera image 1002 , the head becomes dark and blends with the background, so the result of the three-dimensional object detection process 208 is a detection area 1003 that does not include the pedestrian's head.

As shown in (C) of FIG. 10 , in the case of high beam (when the irradiation range of the headlight is set high), since the optical axis is directed upward, even when a pedestrian is located nearby, the head Also included in the irradiation range 1004 . When a pedestrian is observed using the camera image 1005 at this time, as shown in (D) of FIG. 10 , the head is also irradiated with light, and the result of the three-dimensional object detection process 208 is a detection area 1006 including the pedestrian's head.

Therefore, calculation of correction amounts corresponding to low beam and high beam is performed. The method of determining the correction amount is as follows: For example, in the case of low beams, if there is a pedestrian at a position where the head is difficult to be included due to the irradiation range and distance, the amount of the head and body level, for example, 25cm It is set as the correction amount in the upper end direction of the recognition area (correction is performed to expand the upper area of the recognition area). In addition, in the case of high beam, since the head is easily included, the correction amount is about four and a half head and body, for example, 6cm.

The determination of low beam and high beam can also be determined by directly referring to the internal information of the host vehicle, performing image processing on the captured image, and calculating the brightness distribution. In addition, depending on the type of headlights (LED, halogen, etc.) equipped on the vehicle, the value of the correction amount may be calculated in consideration of the diffusion characteristics of these lights. The correction amount may be changed using the result of the external environment determination process 501 . For example, in the external environment judging process 501, when it is judged that the darkness is about dusk, it is judged that the pedestrian's head can be sufficiently detected even if the headlights are low beams, and it is also possible to perform close-up detection compared to nighttime. In the case of light, reduce the adjustment of the correction amount.

In the detection result correction process 505 of the area correction amount calculation unit 402, when it is determined by the detection unit 401 that there is an obstruction in the detection area, the result of the obstruction detection process 502 is used so that the obstruction is not included in the recognition area. The method of calculating the correction amount of the recognition area. For example, the following method can be cited: according to the parallax distribution, the parallax distribution of objects on the near side and objects on the deep side is grouped from the parallax distribution 702 in FIG. method to calculate the correction amount of the recognition area.

In addition, in the detection result correction process 505, the following process may be performed: use not only disparity information, but also use color information, feature value extraction, etc. to obtain the positional relationship between the obstruction 703 and the pedestrian in detail, so as not to include the obstruction 703 703 to calculate the correction amount.

The recognition region correction processing unit 506 of the recognition region correction unit 403 performs size adjustment of the recognition region based on each correction amount calculated by the region correction amount calculation unit 402 . The three-dimensional object detection area is used as the basis of the recognition area, and the correction amount calculated by the distance judgment correction processing 503, the headlight irradiation range judgment correction processing 504, and the detection result correction processing 505 is integrated to correct the final recognition area.

The recognition region correction processing unit 506 performs correction so as to reset, for example, a region corresponding to half the height from the upper part of the detection frame of the three-dimensional object to the ground as the recognition region based on the determination result.

When the object is a pedestrian, the recognition area is corrected so that the upper body including the head of the pedestrian is included in the recognition area, and it is determined whether the object in the corrected recognition area is a pedestrian. Thereby, it is possible to improve the recognition accuracy of whether the person is a pedestrian or not.

FIG. 11 is a diagram showing an example of a recognition region correction processing result.

In (A) of FIG. 11 , for example, in the camera image 1101 in which the feet are cut off during the daytime, the recognition region 11020 is corrected so that the upper body including the pedestrian's head is the recognition region. Area 1102.

Furthermore, in (B) of FIG. 11 , among the camera images 1103 in which the headlights of the own vehicle are at low beam at night, for the camera images 1103 that can be judged that the head of a pedestrian is not included in the detection area, in order to include The upper body of the pedestrian's head is used as the recognition area, and the recognition area 11030 is corrected to be the recognition area 1104 .

In addition, in (C) of FIG. 11 , in the camera image 1105 in which the headlights of the host vehicle are at high beam at night, in order to make the upper body including the pedestrian's head the recognition area, the foot of the pedestrian is cut off. The image 11060 is corrected and used as the recognition area 1106 .

The recognition region correction calculation unit 402 makes the correction amount of the upper region of the recognition region larger than that of the high beam (front headlight) in the case of low beam (when the irradiation range of the headlight illuminating the front of the external environment is set to be low). If the irradiation range of the lamp is set high) the correction is performed in such a way that it becomes larger.

Also, depending on the calculation result of the correction amount of the recognition region, there may be a case where the corrected recognition region coincides with the pre-correction detection region.

As mentioned above, according to one embodiment of the present invention, it is configured that when it is judged that a part of the pedestrian is blocked by an obstruction, the distance between each area and the vehicle is calculated, and the object identified in the immediate area is removed as an obstruction. The image can identify whether it is a pedestrian.

In addition, when it is determined that the foot of the object is cut off, the recognition area is corrected so that the upper body part including the head becomes the recognition area, and when it is determined that the head is not illuminated, the recognition area is corrected to include The recognition region is corrected so that the head region becomes the recognition region, and whether or not it is a pedestrian is determined using the recognition region having the upper body including the head.

Therefore, it is possible to realize an image processing device capable of correctly identifying whether a pedestrian is a pedestrian even when the entire body of a pedestrian existing around the own vehicle cannot be detected.

In the recognition process 404, a plurality of classifiers may be prepared, and the classifiers may be switched according to the correction result of the recognition area.

For example, when the upper body of a pedestrian becomes the recognition area based on the calibration result, instead of using a classifier for the whole body, the classifier created using the learning data of the upper body of the pedestrian is used for recognition.

In addition, the above-mentioned example is a case where the present invention is applied to an image processing device mounted on a vehicle, but the present invention can also be applied to an image processing device mounted on a moving body (article transport vehicle, etc.) other than a vehicle.

In addition, the above example is an example of an image processing device that recognizes a pedestrian (person), but it can also be applied to an image processing device that recognizes a moving object other than a person. For example, it can also be applied to an image processing device that recognizes other vehicles, small animals, and the like.

The aforementioned detection area is an area for detecting an object, and the recognition area is an area for identifying a detected object within the detection area. Therefore, the recognition area often corrects the detection area, but the corrected result may be the same.

Symbol Description

100...stereoscopic camera device, 101...left camera, 102...right camera, 103...image input interface, 104...image processing unit, 105...operation processing unit, 106...storage unit, 107...CAN interface, 108...control processing unit, 109...internal bus, 110...external vehicle network CAN, 203, 204...image data, 205...image processing, 206...image buffer, 207...parallax processing, 208...three-dimensional object detection processing, 209...various object recognition processing, 210...vehicle control processing, 301...three-dimensional object area, 401...detection unit, 402...area correction calculation unit, 403...recognition area correction unit, 404...recognition processing unit, 501...external environment detection processing, 502...occlusion detection processing , 503...distance judgment correction processing, 504...headlight irradiation range judgment correction processing, 505...detection result correction processing, 506...recognition area correction processing, 601, 1002, 1005, 1101, 1103, 1105...camera image, 701, 801, 1003, 1006... detection area, 702... parallax distribution, 703... occlusion, 901... camera shooting angle, 1001... range of light irradiation, 1102, 1104, 1106, 11020, 11030, 11060... recognition area.

Claims (5)

1. An image processing apparatus is characterized by comprising:

an imaging unit that obtains an image by imaging an external environment;

a detection unit that detects image information of the external environment;

a recognition area correction amount calculation unit that calculates a correction amount for recognizing a recognition area of a three-dimensional object based on the image information detected by the detection unit;

an identification region correction unit that corrects the identification region for identifying the image information of the three-dimensional object, based on the correction amount calculated by the identification region correction amount calculation unit; and

a recognition processing unit that performs recognition processing on the three-dimensional object in the recognition area corrected by the recognition area correcting unit,

the detection unit detects a distance from the imaging unit to the three-dimensional object,

the identification area correction unit performs correction for enlarging an upper area of the identification area when the distance detected by the detection unit is equal to or less than a distance set value and when it is determined that the upper portion of the three-dimensional object is not in the identification area,

when the distance detected by the detection unit is equal to or less than a distance set value, the identification area correction unit performs correction so as to reduce the magnitude of correction of the upper area of the identification area as the distance detected by the detection unit increases.

2. The image processing apparatus according to claim 1,

the detection section detects the brightness of the external environment,

the identification region correction amount calculation unit calculates the correction amount of the identification region based on the luminance detected by the detection unit.

3. The image processing apparatus according to claim 1,

the detection unit detects a blocking object that blocks the three-dimensional object,

the identification region correction amount calculation unit calculates the correction amount of the identification region so as to exclude the obstruction detected by the detection unit from the identification region.

4. The image processing apparatus according to claim 2,

the recognition area correction amount calculation unit performs correction such that, when an irradiation range of the headlight irradiated in front of the external environment is set low, a correction amount of an upper area of the recognition area is larger than that when the irradiation range of the headlight is set high.

5. The image processing apparatus according to claim 1,

the three-dimensional object is a pedestrian, and the image processing device is mounted on a vehicle.

Images