JP2009147628A - In-vehicle image processing apparatus and in-vehicle image display device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an in-vehicle image processing apparatus capable of emphasizing an important body in an image in a reduced image and obtaining a less-distorted reduced image, and an in-vehicle image display device that displays the image. <P>SOLUTION: Images of rear parts of both sides of this vehicle are acquired from side rear cameras (S100), and restriction points (the horizon, a straight line extending from a vanishing point, or a lane sectioning line) for not distorting straight lines during image reduction are extracted. Further, restriction points (a vehicle, a vanishing point of a white line, an area of one of infinite points in both rear directions of this vehicle, and its swing suppression direction) for not distorting a swing suppression area in the image are detected (S105). Then an image energy image is generated (S110). Further, Seam obtained by connecting parts where image energy is equal to or smaller than a predetermined value via restriction points set in the image is calculated, and the Seam is deleted to generate the reduced image (S115). Then the calculated reduced image is displayed on a display device (S120). <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to an in-vehicle image processing apparatus and an in-vehicle image display apparatus for reducing and displaying an image while maintaining visibility of important parts around a vehicle.

  Conventionally, when driving a vehicle, an on-board camera is used to acquire an image of the rear side of the vehicle, highlight an object with a relative speed detected by optical flow, etc., and generate an image that is easier for the driver to understand There is a technology that aims to improve safety (see, for example, Patent Document 1).

  On the other hand, there is also a technique for reducing the image while leaving the important part by analyzing the importance of the image and cutting and pasting the part having the high importance (see, for example, Patent Document 2). Furthermore, to improve this, first calculate the image energy image showing the importance, and calculate the line that cuts through the image energy image through the less important part called Seam, and delete that part Have been proposed (see, for example, Non-Patent Document 1).

With these reduced image generation techniques, the surrounding situation can be efficiently displayed on a small monitor. When this is applied to an in-vehicle camera, for example, in the side rear camera, the overtaking vehicle is more emphasized and it is easy to understand whether or not the lane can be changed at a glance.
JP2003-274393 US2007 / 0025637 “Seam Carving for Content-Aware Image Resizing”, S. Avidan (Mitsubishi Electric Research Labs), SIGGRAPH'07

  However, the method described in Non-Patent Document 1 has a problem that the image after reduction is distorted. For example, although the white line of the adjacent lane in the side rear camera image is a straight line, it is interrupted or bent. This occurs when the deleted seam is concentrated in one part of the white line depending on the energy distribution of the image. As a general method, when an edge image is used for an image energy image, roadside structures often have strong edges, and it is easy to run Seaam in the valleys of roadside buildings, and the straight lines nearby there are Often distorted.

  In addition, in the moving image, since the location to be deleted is different for each frame, for example, the appearance of the overtaking vehicle shown in the side rear camera may be blurred from side to side. This occurs because the number of passing vehicles on the right side and the left side of the overtaking vehicle to which the deleted Sea is focused differs from frame to frame.

Thus, in order to apply this method to an in-vehicle moving image, it is necessary to set an appropriate constraint condition for the Seam so as to suppress distortion and shaking.
The present invention has been made in view of these problems, and displays an in-vehicle image processing apparatus that can emphasize an important object in an image in a reduced image and obtain a reduced image with less distortion, and the image. An object is to provide an in-vehicle image display device.

  The in-vehicle image processing apparatus according to claim 1, which has been made to solve such a problem (5: In this section, in order to facilitate understanding of the invention, the “best mode for carrying out the invention is described as necessary. The reference numeral used in the “form” column is attached, but it does not mean that the scope of claims is limited by this reference.) Is an image acquisition means (10), an image energy calculation means (30), and a constraint point arrangement means. (30) and image reduction means (30).

  The image acquisition means (10) is for acquiring images around the host vehicle, and the image energy calculation means (30) calculates and calculates image energy in the image acquired by the image acquisition means (10). A line passing through a portion where the image energy is lower than a predetermined value is calculated.

The restraint point arrangement means (30) sets a restraint point that the line must pass through the image acquired by the image acquisition means (10) regardless of the value of the image energy. The image reduction means (30) The image is reduced by deleting the line portion in the image calculated by the energy calculating means (30).

Further, the image energy calculation means (30) passes the constraint points arranged by the constraint point arrangement means (30) when calculating the line.
The on-vehicle image processing apparatus (5) as described above is an image processing apparatus capable of accurately selecting only an object at a position important for vehicle operation for the driver and performing image reduction and displaying an image that is easy for the driver to see. Become. This will be described below.

  First, “image energy in an image” will be described. This is an index indicating where an important part in an image is. An object that is calculated for each position in the image and creates an energy distribution of the entire image is called an energy image.

  Various methods can be considered for this calculation method, that is, the energy definition method. For example, it can be considered that the edge image, that is, the line density, is a portion where the color change is large. For example, in places that are not important, such as clear sky, line density and screen color change can hardly be made. On the other hand, in places that are considered important in images of urban roads, lanes and other paint are applied to the road surface, there are many vehicles on the road, and buildings are lined up on the side of the road. In addition, because there are advertising billboards lined up on the side of the road and buildings, the line density and color change are large.

  As described above, when the edge extraction processing is performed on the image energy in the image, the portion having a strong edge becomes “high image energy in the image”, while an important one without a pattern, for example, a design with a soft design. The vehicle will be considered unimportant. For this reason, the vehicle is deleted, or the part inside the vehicle is deleted and the appearance of the vehicle is distorted. Therefore, this image energy calculation method is very important.

  According to the “image energy” defined as described above, in the on-vehicle image processing device (5) according to claim 1, the image energy in the image is calculated by the image energy calculation means (30), and the calculated image is calculated. A portion where the energy is lower than a predetermined value is deleted, and the image is reduced.

  As described above, when the image is reduced by deleting a portion where the image energy is lower than a predetermined value, the obtained image is distorted. In particular, the distortion increases when the line passes through the image obliquely and is concentrated at one point.

  Therefore, in the present invention, as described in claim 1, a restriction point through which the line always passes is set. The line that sequentially passes the specified constraint point from one end of the image to the other opposite end is calculated, and by appropriately varying the position of the constraint point, concentration of the line can be prevented, As a result, image distortion during image reduction can be suppressed.

Specifically, as described in claim 2, the constraint point arrangement means (30) may set the constraint points at predetermined intervals in the image acquired by the image acquisition means (10).
In this way, if the constraint points are arranged at predetermined intervals by the constraint point arrangement means (30) in the portions that are not to be distorted, the lines passing there are at the predetermined intervals, so that the lines are not concentrated. . If the lines are not concentrated, the deleted portion does not concentrate, and thus the portion is not distorted when the image is reduced.

  In addition, in the image acquired by the image acquisition means (10), when a straight line such as a horizon, a straight line extending from the vanishing point, or a lane marking is distorted, the driver feels uncomfortable. Therefore, as described in claim 3, the image processing means (10) includes line segment extraction means (30) for extracting a horizon line, a straight line extending from the vanishing point, or a lane division line in the image acquired by the image acquisition means (10), The means (30) may arrange the constraint points on the line segment extracted by the line segment extraction means (30).

  In this way, since the restraint points are arranged at a predetermined interval in the straight line portions, the straight lines are reduced at regular intervals when the image is reduced. Therefore, those straight lines are not distorted.

  By the way, when this reduction by line deletion is applied in a moving image, an object originally located at the same position in the original image may be shaken by the deletion process. For example, in the side rear camera, the other vehicle traveling in the adjacent lane trembles from side to side. This is because the number of lines to be deleted on the right and left sides of the vehicle is not constant every frame. If you want to suppress the shaking in the image, if the number of lines that pass through the left and right is constant, the image is reduced. Can prevent shaking.

  Therefore, as described in claim 4, the apparatus is provided with a rocking suppression area detecting means (30) for detecting the rocking suppression area and the rocking suppression direction in the image acquired by the image acquisition means (10), and is provided with a constraint point arrangement. The means (30) is configured such that the number of lines to be deleted passing through both sides of the fluctuation suppression direction of the fluctuation suppression area in the image detected by the fluctuation suppression area detection means (30) is a predetermined number. It is advisable to arrange the constraint points so that

  In this way, the total number of lines passing between both sides of the oscillation suppression direction of the oscillation suppression area and passing to the image end is, for example, 100 on the left side and 50 on the right side for each frame. The effect of making it constant can be produced. That is, since the position on the image after processing becomes constant in the rocking suppression area, rocking at the time of image reduction can be suppressed.

  Here, the constraint points are not necessarily points, but can be considered all on the line from the end of the oscillation suppression region to the image end. That is, the concept of the constraint point can be extended to a constraint line, and any point on this line can be passed.

  Or you may set the ratio of the total number of the restraint points arrange | positioned on the right and left of a rocking | fluctuation suppression area | region. For example, if 100 points on the left side and 50 points on the right side are set at equal intervals on the white line, distortion and shaking can be suppressed.

  Further, as described above, when the vehicle or the vanishing point of the white line, the infinity point in the backward direction of the host vehicle, or the like swings, the driver feels uncomfortable. Therefore, as described in claim 5, the swing suppression region detecting means (30) is configured to detect the vehicle in the image acquired by the image acquiring means (10), the vanishing point of the white line, the infinity point in the backward direction of the host vehicle. If any region is detected as a swing suppression region, the swing can be suppressed.

  Conventionally, a line in an image starts from the entire upper end of the screen and calculates a path with the shortest image energy toward the lower end of the screen (see Non-Patent Document 1). However, the calculation does not change. In other words, as described in claim 6, when the image energy calculating means (30) calculates the line so that the sum of the image energies on the line passing through the constraint point is minimized, the line to be obtained is the shortest. It becomes a route. When the constrained area is a point, the shortest path from the point to the upper and lower ends is a line to be obtained as it is.

  An image display device (1) according to claim 6 includes an in-vehicle display means (50), an in-vehicle image processing device (5) according to any one of claims 1 to 5, and an in-vehicle image. A display control means (30) for displaying an image reduced by the processing device (5) on the display means (50) is provided.

  According to such an in-vehicle image display device (1), an in-vehicle image display device capable of displaying an image having the characteristics of the in-vehicle image processing device according to any one of claims 1 to 5 ( 1).

  By the way, when the display means (50) is mounted on the host vehicle, various methods are conceivable for the mounting position and the display direction. For example, the display screen of the car navigation device mounted in the front part of the vehicle interior may be used as the display means (50) to display the inside of the vehicle, as described in claim effect 7. The driver may be attached to the rear-viewing mirror portion of the host vehicle so that the driver can visually recognize the display screen.

  In this way, the driver can visually recognize the highlighted object by the same operation as the vehicle rear confirmation operation via the rear confirmation mirror, so it is easy to confirm the presence of the object, and thus driving. Safety at the time can be improved.

  Embodiments to which the present invention is applied will be described below with reference to the drawings. The embodiment of the present invention is not limited to the following embodiment, and can take various forms as long as they belong to the technical scope of the present invention.

  FIG. 1 is a block diagram showing a schematic configuration of an in-vehicle image display device 1 to which the present invention is applied. The vehicle-mounted image display device 1 includes a vehicle-mounted image processing device 5 and a display device 50 as shown in FIG.

The in-vehicle image processing device 5 includes a side rear camera 10, a millimeter wave radar 20, a vehicle speed sensor 40, and an image processing unit 30.
The side rear camera 10 is a small wide-angle CCD camera that can acquire a visible light image for acquiring an image of the surroundings of the host vehicle (not shown), and can be used as a rear-view mirror of the host vehicle so that an image of the rear side of the vehicle can be acquired. It is attached.

  The millimeter wave radar 20 is for detecting the relative speed of an object in the image (in the present embodiment, another vehicle), and is attached to the host vehicle so that millimeter wave radio waves can be transmitted to the rear side of the host vehicle. Yes. The millimeter wave radar 20 transmits a millimeter wave radio wave, receives a reflected wave of the transmitted radio wave, and receives the reflected wave reception intensity, reception azimuth, and the distance from the Doppler shift of the reflected wave to the other vehicle behind the vehicle side. And detect azimuth and relative speed.

  The vehicle speed sensor 40 detects the speed of the host vehicle, and may detect the speed from the number of rotations of the axle, or detects the current position of the host vehicle with a GPS onboard device, and based on the amount of change. It may be one that detects speed.

The image processing unit 30 includes a CPU, ROM, RAM, I / O, and the like (not shown), and executes the following processes (a) to (c).
It comprises a RAM, an I / O, and the like, and executes the following processes (a) to (g).

(A) A horizon line, a straight line extending from the vanishing point, or a lane marking line in the image acquired by the side rear camera 10 is extracted.
(A) In the image, any one of the vehicle, the vanishing point of the white line, and the infinity point in the backward direction of the host vehicle is detected as the swing suppression region and the swing suppression direction.

(C) The image energy in the image acquired by the side rear camera 10 is calculated.
(D) The total number of restraint points to be arranged on the left and right of the rocking suppression direction of the rocking suppression region in the image detected in (a) is determined.

(E) Based on the number of restriction points calculated in (d), restriction points are set at predetermined intervals in the image. At this time, a constraint point is arranged on the line segment extracted in (a) as a constraint point.
(F) A line that sequentially passes the specified restraint point from one end of the image to the other opposite end and passes through a portion where the calculated image energy is lower than a predetermined value is calculated.

(G) The image is reduced by deleting the line portion calculated in (f).
(H) The image reduced by the processing of (g) is displayed on the display device 50.
The display device 50 displays an image reduced by the in-vehicle image processing device 5 and includes a liquid crystal display and an organic EL display. The display device 50 is attached so that the driver can visually recognize the display screen instead of the mirror at the rearview mirror portion of the host vehicle.

  The display device 50 displays an image reduced by the in-vehicle image processing device 5 and includes a liquid crystal display and an organic EL display. The display device 50 is attached so that the driver can visually recognize the display screen instead of the mirror at the rearview mirror portion of the host vehicle.

(Processing in the image processing unit 30)
Next, processing executed by the image processing unit 30 will be described with reference to FIGS. FIG. 2 is a flowchart showing the flow of the main routine of the image processing executed by the image processing unit 30, and FIG. 3 is a flowchart of the constraint point setting process which is a subroutine. FIG. 4 is a flowchart of image energy image generation processing as a subroutine, and FIG. 5 is a flowchart of low image energy location deletion processing as a subroutine.

  In the image processing, as shown in FIG. 2, an image of the rear side of the host vehicle is acquired from the side rear camera 10 in S100, and in the subsequent S105, a constraint point setting process is performed, and a constraint point is included in the image acquired in S100. Is set. The constraint point setting process will be described later.

In subsequent S110, an image energy image of the image acquired in S100 is generated. Details of the image energy image generation processing will be described later.
In subsequent S115, the low image energy location is deleted from the image energy image generated in S110, and a reduced image is obtained. The low image energy location deletion process will be described later.

In subsequent S120, the reduced image obtained in S115 is displayed on the display device 50, the processing is returned to S100, and the image processing is repeated.
(Restriction point setting process)
Next, the constraint point setting process will be described with reference to FIG. In the constraint point setting process, in S200, a constraint point region is extracted from the image acquired in S100 (see FIG. 2). That is, a horizon, a straight line extending from the vanishing point, or a lane division line is extracted from the image by known image processing.

In subsequent S205, any region of the vehicle, the vanishing point of the white line, and the infinity point in the backward direction of the host vehicle is detected as a swing suppression region and a swing suppression direction in the image.
In S210, it is determined in S200 whether a horizon, a straight line extending from the vanishing point, or a lane marking is extracted. If it is determined that one of them has been extracted (S210: Yes), the process proceeds to S220. If none is extracted (S210: No), the process proceeds to S215.

  In S215, it is determined in S205 whether any one of the vehicle, the vanishing point of the white line, and the infinity point in the backward direction of the host vehicle is extracted as the swing suppression region and the swing suppression direction. And when extracted as a rocking | fluctuation suppression area | region and a rocking | fluctuation suppression direction (S215: Yes), a process transfers to S220, and when not extracted (S215: No), a constraint point setting process is complete | finished.

  In S220, a constraint point is set. That is, if there is a line segment extracted in S200, constraint points are arranged on the line segment at a predetermined interval. Further, if there is the swing suppression region and the swing suppression direction detected in S205, the constraint points are arranged on both sides of the swing suppression direction of the swing suppression region so that the total number of constraint points becomes a predetermined number. Is done.

(Image energy image generation processing)
Next, the image energy image generation process will be described with reference to FIG. In the image energy image generation process, a base image energy pattern image is generated in S300. For example, a basic image (base image energy pattern image) serving as a base of the energy image is generated, such as an image having a uniform constant value in the entire region or a constant pattern such as a lattice pattern.

  In S305, an edge image is created for the base image energy pattern image generated in S300. That is, corners and edge portions in the image are extracted to generate an image (edge image). An example of the original image is shown in FIG. 6A and an example of the edge image is shown in FIG.

In S310, the edge image generated in S305 is added to the base image energy pattern image generated in S300. That is, the edge image is superimposed on the base image energy pattern image, and in subsequent S315, the restraint points set in S105 (see FIG. 2) are added to the base image energy pattern image, and the process is terminated.
(Low image energy location deletion processing)
Next, the low image energy location deletion process will be described with reference to FIG. In the low image energy location deletion process, a line to be reduced (hereinafter, this line is referred to as “Seam”) for the image energy image obtained by the image energy image generation process of S110 (see FIG. 2). The number is calculated.

  Specifically, the ratio of the vertical and horizontal lengths of the image size acquired in S100 (see FIG. 2) and the length of the reduced image and the width of the Seam that is deleted from the screen when one Seam is deleted from the screen are deleted. The number of power seams is calculated.

  In subsequent S405, Seam is calculated. Seam is a line that can pass with the minimum energy from the upper end to the lower end of the image, and is obtained by solving the shortest path problem. Specifically, the calculation of Seam secures a memory space (hereinafter referred to as a path image) for storing a path total value having the same size as the image energy image obtained in S240 (see FIG. 3), and 2 of the upper side of the original image. From the line to the bottom line, calculate for all pixels as follows:

  For one pixel in an image, the sum of the smallest value of the path image energy values of the top three pixels and the energy value of the original pixel position of the original image is stored in the path image. As a result, the shortest path value from the upper end of the image to the pixel is stored in the path image, and the smallest thing among the values on the lowest side of the path image is the total energy value of the Seam to be obtained, and the path image is reversed from that point By going back toward the upper end, the shape of the Seam is obtained.

At this time, if there is a restraint point set in S105 (see FIG. 2), even when the image energy at the restraint point is low, the Seam is passed with the highest priority.
In addition, the Seam is calculated, and the image energy value for each Seam is calculated. An example of Seam is shown in FIG. In FIG. 7A, many lines drawn from the upper side to the lower side of the image are Seam.

Subsequently, in S410, the Seam having the smallest image energy is identified from the image energy value of the Seam calculated in S405.
In S415, the Seam portion specified in S405 is deleted from the image energy image, and in subsequent S420, it is determined whether or not the total number of Seams deleted in S415 is equal to or less than the number of Seams to be deleted calculated in S400. The

  If the total number of deleted Seams is equal to or less than the number of Seams to be deleted (S420: Yes), the process proceeds to S405 and the process is repeated. Further, when the total number of deleted Seams is larger than the number of Seams to be deleted (S420: No), the process is terminated. An example of an image that has been subjected to the low image energy location deletion process is shown in FIG.

(Characteristics of in-vehicle image display device 1)
When the image energy is calculated and a portion where the calculated image energy is lower than a predetermined value is deleted, the image is reduced. However, if the image is reduced by deleting a portion where the image energy is lower than a predetermined value in this way, the resulting image is distorted. In particular, the distortion increases when the line passes through the image obliquely and is concentrated at one point.

  However, in the in-vehicle image display device 1, a location where the calculated image energy is lower than a predetermined value is specified as a constraint point, and a Seam that sequentially passes the specified constraint point from one end of the image to the other opposite end is calculated. Therefore, the concentration of Seam can be prevented, and as a result, distortion of the image at the time of image reduction can be suppressed.

  Further, when the seam is concentrated, the seam of the portion is deleted, so that the distortion of the screen increases. However, in the in-vehicle image display device 1, the restraint points are set at predetermined intervals in the image. ing. That is, for the portions that are not to be distorted, constraint points are arranged at a predetermined interval.

  Therefore, the Seam passing there is a predetermined interval, and the Seam is not concentrated. Since the portion to be deleted does not concentrate unless the Seaam is concentrated, the portion is not distorted when the image is reduced.

  In particular, restraint points are arranged at equal intervals in areas where the driver feels uncomfortable when there is distortion, such as straight lines extending from the horizon, vanishing points, or lane markings. Sometimes these straight lines are reduced at regular intervals. Therefore, distortion of those straight lines can be suppressed.

  In addition, parts such as the vanishing point of the vehicle or the white line and the infinity point in the backward direction of the host vehicle that make the driver feel uncomfortable are detected, and the fluctuation of the fluctuation suppression area in the detected image is detected. Restriction points are arranged by controlling the total number of seams passing through both sides in the suppression direction, for example, 100 on the left side and 50 on the right side.

Therefore, since the position on the image after processing is constant in the vibration suppression area, it is possible to suppress the vibration when the image is reduced.
In addition, the Seam is specified so that the total image energy on the Seam passing through the constraint point is minimized. Therefore, since the Seam is the shortest path, an extra portion in the image is not deleted when the image is reduced, so that distortion of the image can be suppressed.

  In addition, the display device 50 is attached to the rear-viewing mirror portion of the host vehicle so that the driver can visually recognize the display screen. Therefore, the driver can visually recognize the highlighted object by the same operation as the vehicle rearward confirmation operation through the rearward confirmation mirror, so it is easy to confirm the presence of the object, and thus safety during driving. Can be improved.

[Other Embodiments]
As mentioned above, although embodiment of this invention was described, this invention is not limited to this embodiment, A various aspect can be taken.

  (1) In the above embodiment, the display device 50 is attached to the rearview mirror portion of the host vehicle so that the driver can visually recognize the display screen, but is attached to the front part of the passenger compartment of the host vehicle. Display may be performed on the inside of the vehicle using the display screen of the car navigation device.

  (2) In the above embodiment, an edge image is generated to generate an image energy image (see S305 in FIG. 4). However, instead of using an edge, another function is used in accordance with the specification of the reduction target image. It may be used. Specifically, functions such as an edge square function, a saliency function, a Harris corner function, a gaze recognition result function, a face recognition result function, an entropy function, a segmentation function, or a HoG (abbreviation of Histogram of Gradients) function may be used. .

  (3) The side rear camera 10 may be an infrared camera instead of the visible light camera. An infrared camera is effective for highlighting people or bicycles in a reduced image when the surroundings of the vehicle are dark, such as at night.

  (4) In the above embodiment, the Seam direction is assumed to be up and down so as to reduce the horizontal width of the image. However, when the vertical width is reduced, the Seam direction can be replaced with left and right.

1 is a block diagram illustrating a schematic configuration of an in-vehicle image display device 1. FIG. It is a flowchart which shows the flow of a process of the main routine of an image process. It is a flowchart of the constraint point setting process which is a subroutine. It is a flowchart of the image energy image generation process which is a subroutine. It is a flowchart of the low image energy location deletion process which is a subroutine. It is a figure which shows the example of the edge image produced | generated in an image energy image production | generation process. It is a figure which shows the example of the image in which the low image energy location deletion process was made.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 1 ... Car-mounted image display apparatus, 5 ... Car-mounted image processing apparatus, 10 ... Side rear camera, 20 ... Millimeter wave radar, 30 ... Image processing part, 40 ... Vehicle speed sensor, 50 ... Display apparatus.

Claims (8)

Image acquisition means for acquiring an image around the host vehicle;
Image energy calculating means for calculating image energy in the image acquired by the image acquiring means and calculating a line passing through a portion where the calculated image energy is lower than a predetermined value;
Constraint point arrangement means for setting a constraint point that the line must pass through regardless of the energy value in the image acquired by the image acquisition means;
Image reduction means for reducing the image by deleting a line portion in the image calculated by the image energy calculation means;
With
The image energy calculating means includes
An on-vehicle image processing apparatus characterized in that when the line is calculated, the restriction point arranged by the restriction point arrangement unit is passed through the line. The in-vehicle image processing apparatus according to claim 1,
The restraint point arrangement means includes:
An in-vehicle image processing apparatus, wherein the restraint points are set at predetermined intervals in an image acquired by the image acquisition means. The in-vehicle image processing apparatus according to claim 2,
A horizon in the image acquired by the image acquisition means, a straight line extending from the vanishing point or a line segment extraction means for extracting a lane division line,
The restraint point arrangement means includes:
An in-vehicle image processing apparatus, wherein the restraint points are arranged on a line segment extracted by the line segment extraction means. An in-vehicle image processing apparatus according to any one of claims 1 to 3,
A swing suppression region detection means for detecting a swing suppression region and a swing suppression direction in the image acquired by the image acquisition means;
The restraint point arrangement means includes:
The restraint points so that the number of lines to be deleted that pass through both sides of the swing suppression direction of the swing suppression region in the image detected by the swing suppression region detection means becomes a predetermined number. An in-vehicle image processing apparatus characterized by comprising: The in-vehicle image processing apparatus according to claim 4,
The swing suppression region detecting means is
An in-vehicle image processing apparatus that detects any region of a vehicle, a vanishing point of a white line, and an infinite point in the backward direction of the host vehicle as an oscillation suppression region in the image acquired by the image acquisition unit. In the in-vehicle image processing apparatus according to any one of claims 1 to 5,
The image energy calculating means includes
An in-vehicle image processing apparatus that calculates a line so that a total of image energies on the line passing through the constraint point is minimized. In-vehicle display means;
The in-vehicle image processing device according to any one of claims 1 to 6,
Display control means for displaying on the display means an image reduced by the in-vehicle image processing apparatus;
An in-vehicle image display device comprising: The in-vehicle image display device according to claim 7,
The display means includes
An in-vehicle image display device, wherein the driver is attached to a mirror portion for visually recognizing the rear of the host vehicle so that a driver can visually recognize a display screen.