Vehicle lamp and its control device and control method

Abstract

A vehicle lamp capable of forming a light distribution that reduces the influence of noise is provided. [Solution] A control unit 220, which is a control device for a vehicle lamp 200, includes an image processing unit 221. The image processing unit 221 receives primary image data IMG1 captured by a camera unit 210, generates intermediate image data IMG2 corresponding to the primary image data IMG1, and generates illuminance control data that defines the illuminance distribution of the vehicle lamp 200 based on the intermediate image data IMG2. For pixels in the primary image data IMG1 whose pixel values ​​fall within a predetermined range, the effective number of gradations in the intermediate image data IMG2 is smaller than the number of gradations in the primary image data IMG1.

Description

Technical Field

[0001] The present invention relates to a vehicle lamp.

Background Art

[0002] Vehicle lamps play an important role in safe driving at night or in tunnels. If the vehicle in front is brightly illuminated over a wide range, giving priority to visibility by the driver, there is a problem that glare is given to the driver of a preceding vehicle or an oncoming vehicle (hereinafter referred to as a front vehicle) or a pedestrian existing in front of the own vehicle.

[0003] In recent years, based on the state around the vehicle, ADB (Adaptive Driving Beam) technology has been proposed to dynamically and adaptively control the light distribution pattern. The ADB technology detects the presence or absence of a front vehicle or a pedestrian, and reduces the glare given to the driver of the front vehicle or the pedestrian by dimming or turning off the area corresponding to the vehicle or the pedestrian.

Prior Art Documents

Patent Documents

[0004]

Patent Document 1

Patent Document 2

Patent Document 3

Summary of the Invention

Problems to be Solved by the Invention

[0005] The present invention has been made in view of such circumstances, and an exemplary object of one aspect thereof is to provide a vehicle lamp capable of forming a light distribution pattern excellent in visibility. Another object is to provide a vehicle lamp capable of forming a light distribution pattern that is easy for an in-vehicle camera to recognize a target.

[0006] In addition, the present inventors studied a vehicle lamp that captures an image in front of the vehicle with a camera and controls the light distribution based on the captured image data, and came to recognize the following problems. Usually, since vehicle lamps are mainly used at night, the object to be imaged by the camera is dark.

[0007] In order to capture a dark field of view with sufficient brightness for object recognition, it is necessary to increase the sensitivity of the camera. However, increasing the sensitivity reduces the signal-to-noise ratio. Controlling the light distribution based on image data with a low signal-to-noise ratio may cause the noise component to affect the projected light distribution pattern, resulting in a decrease in visibility. This problem should not be regarded as a common knowledge of those skilled in the art.

[0008] The present invention has been made in such a situation, and an exemplary object of one aspect thereof is to provide a vehicle lamp capable of forming a light distribution with reduced noise influence.

Means for Solving the Problems

[0009] One aspect of the present invention relates to a vehicle lamp. The vehicle lamp can generate a light distribution pattern based on an image captured by a camera, and a plurality of control modes with different combinations of the spatial resolution of the light distribution pattern and the update speed of the light distribution pattern are switchable.

[0010] One aspect of the present invention relates to a control device for a vehicle lamp or a vehicle lamp. The control device includes an image processing unit that receives initial image data captured by a camera, generates intermediate image data according to the initial image data, and generates illuminance control data that defines the illuminance distribution of the vehicle lamp based on the intermediate image data. For pixels whose pixel values are included in a predetermined range among the pixels of the initial image data, the effective number of gradations in the intermediate image data is smaller than the number of gradations in the initial image data.

[0011] In addition, any combination of the above components, and those obtained by converting the expression of the present invention among methods, devices, systems, etc. are also effective as aspects of the present invention.

Effects of the Invention

[0012] According to one aspect of the present invention, a light distribution pattern excellent in visibility can be formed, or a light distribution pattern easy for an in-vehicle camera to recognize an object can be formed. Further, according to another aspect of the present invention, a light distribution with reduced influence of noise can be formed.

Brief Description of Drawings

[0013]

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

Figure 11

Figure 12

Figure 13

Figure 14

Figure 15

Figure 16

BEST MODE FOR CARRYING OUT THE INVENTION

[0014] First, the outline of a vehicle lamp according to several representative embodiments will be described. 1. One embodiment of the present invention relates to a vehicle lamp. The vehicle lamp can generate a light distribution pattern based on an image captured by a camera, and a plurality of control modes with different combinations of the spatial resolution of the light distribution pattern and the update speed of the light distribution pattern are switchable.

[0015] By performing appropriate signal processing according to the driving environment and the user's preference, an appropriate light distribution pattern can be generated.

[0016] In one embodiment, the vehicle lamp may include a light distribution pattern generator that generates a light distribution pattern based on an image captured by a camera, a light source unit that irradiates the front of the vehicle according to the light distribution pattern, and a mode controller that adaptively controls the combination of the spatial resolution of the light distribution pattern and the update speed of the light distribution pattern.

[0017] The plurality of control modes may include a first mode that generates a light distribution pattern with relatively high resolution and low speed, and a second mode that generates a light distribution pattern with relatively low resolution and high speed. As a result, expensive hardware capable of high-speed processing is not required. Alternatively, the human eye has the characteristic that it can recognize a fast-moving object only with low resolution and a slow-moving object with high resolution, and a light distribution pattern that matches the characteristics of the human eye can be generated.

[0018] The control mode may be adaptively selected according to the driving environment. Alternatively, the control mode may be manually selected by the driver.

[0019] The control mode may be adaptively selected according to the ambient illuminance. The characteristics of the human eye change between bright and dark environments. Therefore, by considering the ambient illuminance, a high-visibility light distribution pattern can be provided. Alternatively, under the condition of constant sensitivity, the exposure time per frame of the camera can vary between bright and dark environments. Therefore, by considering the ambient illuminance, an operation corresponding to the operation of the camera can be provided.

[0020] The ambient illuminance may be detected based on an image captured by the camera. By also using the camera as an illuminance sensor, hardware can be reduced.

[0021] The control mode may be adaptively selected according to the driving speed. The driving speed may be used as an indicator of the moving speed of the irradiated object (irradiation target).

[0022] The control mode may be adaptively selected according to the speed of the irradiated object (irradiation target).

[0023] The control mode may be adaptively selected based on the spatial frequency of an image captured by the camera. The control mode may be adaptively selected based on the moving speed of an object included in an image captured by the camera.

[0024] The control mode may be selected based on the type of road on which the vehicle is traveling.

[0025] An image captured by the camera may be divided into a plurality of sub-regions, and the control mode may be set for each sub-region. It may be divided into a relatively bright sub-region and a relatively dark sub-region. Alternatively, it may be divided into a sub-region with a relatively fast displacement speed of the target and a sub-region with a relatively slow displacement speed.

[0026] 2. One embodiment of the present invention relates to a control device for a vehicle lamp. The control device receives initial image data captured by a camera, generates intermediate image data according to the initial image data, and includes an image processing unit that generates illuminance control data for defining the illuminance distribution (light distribution pattern) of the vehicle lamp based on the intermediate image data. For pixels among the pixels of the initial image data whose pixel values are included in a predetermined range, the effective number of gradations in the intermediate image data is smaller than the number of gradations in the initial image data.

[0027] The light distribution pattern is treated as a set of a plurality of individual regions (meshes), and the illuminance of each mesh is determined based on the pixel value of the corresponding position included in the initial image data. Here, when noise is superimposed on the pixel value, if noise is included in a specific range of pixel values, the influence of the noise may become prominent in the light distribution pattern. Therefore, for pixels among the pixels included in the initial image data where the influence of noise becomes prominent in the final illuminance distribution, by reducing the effective number of gradations, a light distribution with reduced influence of camera-derived noise can be formed. By reducing the number of gradations only within a predetermined range, the original high number of gradations can be maintained for pixels where noise is less likely to be noticeable.

[0028] The predetermined range may be from zero to a predetermined upper limit value. Alternatively, the predetermined range may be from a predetermined lower limit value to the maximum gradation value. There may be a plurality of predetermined ranges.

[0029] The process of generating the illuminance control data may include a process of comparing the pixel values of the intermediate image data or the image data obtained therefrom with a threshold value. In this case, by defining the predetermined range in association with the threshold value, the influence of noise can be reduced.

[0030] The process of generating the illuminance control data may include a process of changing the contrast of the intermediate image data or the image data obtained therefrom. By reducing (or increasing) the contrast, the noise in the dark part may be amplified. Therefore, the influence of noise can be reduced by reducing the number of gradations in a predetermined range associated with the dark part.

[0031] The process of generating the illuminance control data may include a process of inverting the pixel values of the intermediate image data or the image data obtained therefrom in tone. In this case, among the intermediate image data, the darker the area, the higher the illuminance, so the noise included in the dark area is emphasized in the bright area of the illuminance distribution. Therefore, the influence of noise can be reduced by associating a predetermined range with the dark area and reducing the number of tones.

[0032] The image processing unit may compare the pixel values of the intermediate image data with a threshold value and generate illuminance control data based on the comparison result. When paying attention to a plurality of adjacent pixels, if the pixel value including noise exceeds or does not exceed the threshold value, the influence of noise on the illuminance control data becomes significant. In this case, by reducing the number of tones of the intermediate image data, the influence of noise can be suppressed.

[0033] The process of generating the intermediate image data may include a process of multiplying the pixel values of the initial image data by a coefficient smaller than 1.

[0034] The process of generating the intermediate image data may include a process of rounding the lower N bits (an integer where N ≥ 1) of the pixel values of the initial image data.

[0035] The image processing unit may vary the degree of reducing the number of tones according to the amplitude of the noise included in the initial image data. Since the amount of noise may change depending on the environment such as temperature, the influence of noise can be appropriately reduced according to the situation.

[0036] The above is an overview of the vehicle lamp. Hereinafter, the present invention will be described with reference to the drawings based on preferred embodiments. The embodiments are illustrative and not restrictive of the invention, and all features and combinations thereof described in the embodiments are not necessarily essential to the invention. The same or equivalent components, members, and processes shown in each drawing are denoted by the same reference numerals, and redundant descriptions are omitted as appropriate. Also, the scales and shapes of the respective parts shown in each figure are set for convenience to facilitate the explanation, and are not to be construed restrictively unless otherwise specified. Further, when terms such as "first" and "second" are used in this specification or claims, these terms do not represent any order or importance, but are for distinguishing one configuration from another.

[0037] FIG. 1 is a block diagram of a vehicle lamp system according to a first embodiment. The vehicle lamp system 100 includes a vehicle lamp 200, a vehicle ECU (Electronic Control Unit) 102, and a battery 104. The battery voltage VBAT generated by the battery 104 is supplied as a power supply voltage to the vehicle lamp 200. The vehicle ECU 102 and the vehicle lamp 200 comply with protocols such as CAN (Controller Area Network) and LIN (Local Interconnect Network), and are capable of communicating. The vehicle ECU 102 controls the lighting and extinguishing of the vehicle lamp 200, and also transmits vehicle information such as vehicle speed and steering to the vehicle lamp 200. A fail signal or the like can be transmitted from the vehicle lamp 200 to the vehicle ECU 102.

[0038] The vehicle lamp 200 is configured to dynamically and adaptively control the light distribution pattern PTN based on an image captured by the camera unit 210 (hereinafter referred to as the camera image IMG). Note that this camera image IMG corresponds to one frame of data constituting a moving image. The camera unit 210 may include only one camera, or may include a plurality of cameras with different resolutions and / or frame rates. When the camera unit 210 includes a plurality of cameras, the camera image IMG is understood as a collective term for the output data of the plurality of cameras.

[0039] The light distribution pattern PTN is understood as the two-dimensional illuminance distribution of the irradiation pattern 902 formed by the vehicle lamp 200 on the virtual vertical screen 900 in front of the vehicle. The vehicle lamp 200 can adaptively switch between a plurality of control modes with respect to the generation of the light distribution pattern PTN. The plurality of control modes differ in the combination of the spatial resolution of the light distribution pattern PTN and the update speed (switching speed) of the light distribution pattern.

[0040] The light distribution pattern PTN is divided into a plurality of meshes (regions), and the illuminance of the same mesh is constant. The spatial resolution of the light distribution pattern PTN is associated with the fineness (coarseness) of the mesh.

[0041] Figs. 2(a) to (c) are diagrams for explaining the resolution of the light distribution pattern PTN. Fig. 2(a) shows an example of the camera image IMG, and Figs. 2(b) and (c) are diagrams showing an example of the light distribution pattern PTN corresponding to the camera image IMG of Fig. 2(a). Here, for the sake of easy explanation, an example will be described in which a specific object is extracted from the camera image IMG, and the mesh at the location where the specific object exists is shaded (i.e., the illuminance is zero). The specific object is an object that should not be dazzling, and the preceding vehicle 904, the oncoming vehicle 906, and the pedestrian 908 are exemplified. In Figs. 2(b) and (c), the shaded areas are hatched. The light distribution pattern PTN in Fig. 2(b) has a higher resolution than the light distribution pattern in Fig. 2(c). A higher resolution can achieve shading that more accurately conforms to the shape of the specific object.

[0042] If the resolution of the light distribution pattern PTN is further increased, it is possible to shield only the rear window portion of the preceding vehicle 904 or only the front window portion of the oncoming vehicle 906, and actively increase the illuminance of the body. Similarly, it is possible to shield only the face portion of the pedestrian 908 and actively increase the illuminance of the body portion. This can further enhance the visibility of the driver of the host vehicle while preventing glare.

[0043] Returning to FIG. 1, the specific configuration of the vehicle lamp 200 will be described. The vehicle lamp 200 includes a control unit 220 and a light source unit 230 in addition to the camera unit 210. In FIG. 1, the camera unit 210 is built into the vehicle lamp 200, but it may be provided on the vehicle side.

[0044] The control unit 220 includes a light distribution pattern generator 222 and a mode controller 224. The control unit 220 is also referred to as the lamp ECU. The light distribution pattern generator 222 generates a light distribution pattern PTN based on the camera image IMG. The control unit 220 can be composed of a digital processor, for example, a combination of a CPU, a microcomputer, and a software program, or it can be composed of an FPGA (Field Programmable Gate Array), an ASIC (Application Specified IC), etc.

[0045] The light source unit 230 receives data instructing the light distribution pattern PTN from the light distribution pattern generator 222 and is configured to be able to form an illuminance distribution corresponding to the light distribution pattern PTN in front of the vehicle. The configuration of the light source unit 230 is not particularly limited. For example, it may include a semiconductor light source such as an LD (laser diode) or an LED (light emitting diode), and a lighting circuit for driving the semiconductor light source to light up. The light source unit 230 may include a matrix-type pattern forming device such as a DMD (Digital Mirror Device) or a liquid crystal device for forming an illuminance distribution corresponding to the light distribution pattern PTN.

[0046] The mode controller 224 adaptively controls the combination of the spatial resolution of the light distribution pattern PTN and the update speed (frame rate) of the light distribution pattern PTN, that is, the control mode of the light distribution pattern generator 222.

[0047] The above is the basic configuration of the vehicle lamp 200. Subsequently, its operation will be described.

[0048] Figs. 3(a) and (b) are diagrams for explaining the generation of the light distribution pattern PTN in different control modes. The update period Ts1 of the light distribution pattern PTN in the control mode shown in Fig. 3(a) is shorter than the update period Ts2 of the control mode shown in Fig. 3(b). The resolution of the light distribution pattern PTN in the control mode shown in Fig. 3(a) is lower than the resolution of the control mode shown in Fig. 3(b).

[0049] Figs. 4(a) and (b) are time charts of the light distribution control of the vehicle lamp 200 in different control modes. Fig. 4(a) shows a control mode that prioritizes resolution. The camera image IMG is generated every predetermined frame period TF. The control unit 220 receives the camera image IMG once every multiple frames, and generates the light distribution pattern PTN by image processing based on the received camera image IMG. The processing time TP required to generate one high-resolution light distribution pattern PTN from one camera image IMG is longer than the frame period TF. The update period TS of the light distribution pattern PTN is subject to the constraint of the processing time TP.

[0050] Fig. 4(b) shows a control mode that prioritizes update speed. The control unit 220 receives the camera image IMG every frame, and generates the light distribution pattern PTN by image processing based on the received camera image IMG. The processing time TP required to generate one low-resolution light distribution pattern PTN from one camera image IMG is shorter than the frame period TF.

[0051] FIG. 5 is a diagram for explaining a plurality of control modes in an embodiment. The horizontal axis represents the update speed (reciprocal of the update period) of the light distribution pattern PTN, and the vertical axis represents the resolution. The light distribution pattern generator 222 can switch at least between a first mode MODE1 and a second mode MODE2. In the first mode MODE1, the light distribution pattern PTN is generated at a relatively high resolution and a low speed, and in the second mode MODE2, the light distribution pattern PTN is generated at a relatively low resolution and a high speed.

[0052] As described above, the control unit 220 can be implemented by a processor such as a microcomputer or a CPU. The calculation amount of the light distribution pattern generator 222 increases as the resolution of the light distribution pattern PTN increases, and also increases as the update speed of the light distribution pattern PTN increases.

[0053] In FIG. 5, in order to generate the light distribution pattern PTN in the region 910 corresponding to the highest resolution and the highest update speed, the calculation power required for the processor becomes very high. Generally, a high-speed processor is expensive, and it is often difficult to mount it on a vehicle lamp from the perspective of cost. Conversely, it can be said that when the calculation power of the processor is not so high, the resolution and the update speed are in a trade-off relationship. In the presence of such hardware constraints, the operation region of the light distribution pattern generator 222 may be limited to the region 912 in FIG. 5. This can reduce the calculation power required for the processor, so that an expensive CPU or the like is not required, and an inexpensive processor can be adopted.

[0054] Inside the region 912, in addition to the first mode MODE1 and the second mode MODE2, an intermediate third mode MODE3 may be supported. Furthermore, a fourth mode MODE4, which is a combination of low resolution and low update speed, may be supported.

[0055] By supporting the first mode MODE1 and the second mode MODE2, the following effects can also be enjoyed. The human eye has the characteristic that it can only recognize objects moving at high speed with low resolution and objects moving at low speed with high resolution. Therefore, by making it possible to switch between the first mode MODE1 and the second mode MODE2, a light distribution pattern that matches the biological characteristics of the human eye can be generated.

[0056] Subsequently, the switching of the control mode will be described. It is desirable to adaptively select the control mode according to the driving environment. Specifically, it can be selected based on at least one of the following three parameters, or it can be selected by comprehensively considering multiple parameters.

[0057] 1. Ambient brightness (environmental illuminance) The first parameter is the ambient brightness. Even in an environment where the headlight should be turned on, the ambient brightness varies greatly. For example, in the evening or early morning, it is brighter than at night. Also, even at night, an urban area with many streetlights is brighter than a suburban area with few streetlights. Also, even inside a tunnel, the brightness varies greatly depending on the number and brightness of the lights.

[0058] The human eye has rod cells with low resolution but high sensitivity and cone cells with high resolution but low sensitivity. It is known that cone cells are activated in bright environments and rod cells are activated in dark environments. That is to say, in a dark environment, the resolution of the human eye decreases, so even if the light distribution pattern PTN is controlled with high resolution, it can be said that it cannot be recognized.

[0059] Therefore, the mode controller 224 may select a control mode with higher resolution as the ambient brightness increases, and select a control mode with lower resolution as it gets darker. When the first mode MODE1 to the third mode MODE3 in FIG. 5 can be selected, the first mode MODE1 may be selected in a bright ambient situation, and the third mode MODE3 and the second mode MODE2 may be selected as it gets darker.

[0060] To measure the ambient brightness, as shown in FIG. 1, an illuminance sensor 240 for measuring the ambient illuminance may be provided in the vehicle lamp 200. The mode controller 224 may select a control mode having an appropriate resolution according to the ambient illuminance.

[0061] Instead of the illuminance sensor 240, the camera unit 210 may be used as an illuminance sensor. The camera image IMG includes information on the ambient illuminance. Therefore, the mode controller 224 may estimate the ambient illuminance by performing arithmetic processing on the camera image IMG. For example, the ambient illuminance may be estimated by taking the average of the values (pixel values) of a plurality of pixels in the camera image IMG. Alternatively, the pixel values of the area in the camera image IMG that is not irradiated by the emitted light of the vehicle lamp 200 may be extracted, and the ambient illuminance may be estimated from the pixel values. By estimating the ambient illuminance from the camera image IMG, the illuminance sensor can be omitted.

[0062] 2. Relative speed between the irradiated object and the host vehicle The second parameter is the relative speed between the irradiated object (target) and the host vehicle. The irradiated objects here include vehicles, road signs, pedestrians, road surfaces, delineators, street lights, etc. In other words, the control mode may be selected according to the time frequency of the field of view in front of the host vehicle. That is, when the relative speed of the irradiated object with respect to the host vehicle is high, a control mode with a high update speed may be selected, and as the relative speed becomes low, a control mode with a low update speed may be selected. When the first mode MODE1 to the third mode MODE3 in FIG. 5 are selectable, the second mode MODE2 may be selected in a situation where the relative speed is high, and as it slows down, the third mode MODE3 and the first mode MODE1 may be selected.

[0063] For example, the mode controller 224 may switch the control mode based on the displacement speed of the irradiated object included in the camera image IMG. This enables control that reflects the relative speed between the irradiated object and the host vehicle.

[0064] Alternatively, when the traveling speed of the bicycle is high, the relative speed tends to be high, and when the traveling speed of the bicycle is low, the relative speed tends to be low. Therefore, the light distribution pattern generator 222 may switch the control mode based on the traveling speed of the bicycle.

[0065] 3. Shape and arrangement of the irradiated object The third parameter is the fineness of the shape and arrangement of the irradiated object (or target), in other words, the spatial frequency of the field of view in front of the host vehicle. When the spatial frequency of the field of view is high, a control mode with high resolution may be selected, and as the spatial frequency decreases, a control mode with low resolution may be selected. When the first mode MODE1 to the third mode MODE3 in FIG. 5 can be selected, the first mode MODE1 may be selected in a situation where the spatial frequency is high, and the third mode MODE3 and the second mode MODE2 may be selected as it decreases.

[0066] The spatial frequency of the field of view in front of the host vehicle may be calculated from the camera image IMG. The mode controller 224 may perform Fourier analysis on the image data and calculate the spatial frequency.

[0067] The mode controller 224 may directly or indirectly acquire each of the first to third parameters and select a control mode, but the first to third parameters can also be estimated from the driving environment. Therefore, the mode controller 224 may select a control mode based on the driving environment.

[0068] In one embodiment, the control mode may be selected according to the type of road on which the host vehicle is traveling. The type of road may be classified into urban areas, suburbs, highways, tunnels, etc. The determination of the type of road may be made based on information from the car navigation system, or based on vehicle information such as vehicle speed and steering, or based on an image captured by the camera unit 210.

[0069] Focusing on the first parameter, since the urban area is relatively bright and objects can be recognized with high resolution, the first mode MODE1 or the third mode MODE3 may be selected. On the contrary, since the suburbs are relatively dark and the human eye can only recognize objects with low resolution, the second mode MODE2 or the third mode MODE3 may be selected.

[0070] Focusing on the second parameter, in the urban area, the driving speed is slow and the relative speed between the irradiated object and the vehicle tends to be slow. Therefore, the first mode MODE1 or the third mode MODE3 may be selected. On the contrary, on highways and in the suburbs, the driving speed is fast and the relative speed between the irradiated object and the vehicle tends to be fast. Therefore, the second mode MODE2 or the third mode MODE3 may be selected.

[0071] Focusing on the third parameter, in the urban area, there may be many relatively small targets such as pedestrians and road signs, so the spatial frequency tends to be high. Therefore, the first mode MODE1 or the third mode MODE3 may be selected. On the other hand, on highways and in the suburbs, the number of pedestrians and road signs is relatively small, so the spatial frequency tends to be low. Therefore, the second mode MODE2 or the third mode MODE3 may be selected.

[0072] So far, the case of generating one light distribution pattern in the same control mode has been described. However, it is not limited to this, and one camera image IMG may be divided into a plurality of sub-regions, and the optimal control mode may be selected for each sub-region.

[0073] Figures 6(a) and (b) are diagrams showing a camera image IMG divided into a plurality of sub-regions. In Fig. 6(a), the camera image IMG is divided into two parts vertically. In Fig. 6(b), the camera image IMG is divided into five parts: the center, top, bottom, left, and right.

[0074] For each sub-region, the tendencies of the brightness (first parameter), temporal frequency (second parameter), and spatial frequency (third parameter) of the field of view in front of the vehicle are quite different. Therefore, by dividing into a plurality of sub-regions, more appropriate control becomes possible.

[0075] In the division pattern of Fig. 6(a), since the upper sub-region SR1 may include an irradiated object farther away than the lower sub-region SR1, the displacement speed is slow, that is, the time frequency tends to be low. Also, since a distant object appears smaller than a nearby object, the spatial frequency tends to be high. Therefore, in the upper sub-region SR1, a mode that prioritizes resolution may be used, and in the lower sub-region SR2, a mode that prioritizes update speed may be used.

[0076] In the division pattern of Fig. 6(b), the central sub-region SR1 includes the vanishing point and may include an irradiated object farther away than the other sub-regions SR2 to SR5. Therefore, the displacement speed is slow, that is, the time frequency tends to be low, and the spatial frequency tends to be high. Thus, the sub-region SR1 may use a mode that prioritizes resolution.

[0077] On the other hand, in the left and right sub-regions SR2 and SR3, there is a high possibility that an oncoming vehicle or a vehicle overtaking the host vehicle will appear, and these targets tend to displace at high speed. Therefore, in the sub-regions SR2 and SR3, a mode that prioritizes update speed may be used.

[0078] Note that the division pattern may be adaptively switched according to the type of the driving road, the vehicle speed of the host vehicle, etc.

[0079] Subsequently, the image processing in the light distribution pattern generator 222 will be described. Most simply, the light distribution pattern generator 222 may detect a specific target based on the camera image IMG and perform control to block the portion of the detected target.

[0080] More advancedly, the light distribution pattern generator 222 may change the illuminance of each mesh of the light distribution pattern PTN based on the value (pixel value) of the pixel of the camera image IMG corresponding to that mesh. This is called contrast control. Figs. 6(a) and (b) are diagrams for explaining contrast control. The horizontal axis represents the pixel value of the camera image, and the vertical axis represents the illuminance of the corresponding mesh.

[0081] Figure 7(a) is a diagram for explaining low contrast control. In low contrast control, according to the brightness of the irradiated object, in other words, according to the pixel value, the illuminance is set lower for brighter objects and higher for darker objects. As a result, the brightness difference between the dark part and the bright part becomes smaller, and in particular, the visibility of the dark part can be enhanced. In addition to the solid line of the right-sloping line, stepped discrete control as shown by the dashed line or control along a curve as shown by the dash-dotted line may be performed. Alternatively, the illuminance of the mesh may be adjusted by feedback control so that the pixel value of each mesh approaches a predetermined target value.

[0082] Figure 7(b) is a diagram for explaining high contrast control. In high contrast control, according to the brightness of the irradiated object, the illuminance is set higher for brighter objects and lower for darker objects. As a result, the brightness difference between the dark part and the bright part becomes larger. A visual field with a large brightness difference has the advantage that the human eye can easily recognize the position and shape of an object instantaneously. In addition to the control based on the right-ascending line, stepped control as shown by the dashed line or control along a curve as shown by the dash-dotted line may be performed.

[0083] Next, a configuration example of the light source unit 230 will be described. FIG. 8 is a diagram showing a configuration example of the light source unit 230. The light source unit 230A in FIG. 8 includes a light source 232, a lighting circuit 234, and a patterning device 236. In addition, the light source unit 230A may include a reflection optical system or a transmission optical system (not shown).

[0084] As the light source 232, a high-brightness semiconductor light source such as an LED or an LD is suitable. The lighting circuit 234 supplies a stabilized drive current (lamp current) to the light source 232 and causes the light source 232 to emit light at a predetermined luminance. The emitted light of the light source 232 is incident on the patterning device 236.

[0085] The patterning device 236 can use a DMD or a liquid crystal panel. The DMD is an array of micromirrors with individually controllable reflection angles, and the effective reflectivity can be controlled in multiple gradations for each micromirror. The liquid crystal panel is an array of pixels with individually controllable transmittances, and the transmittance can be controlled in multiple gradations for each pixel.

[0086] FIG. 9 is a diagram showing another configuration example of the light source unit 230. The light source unit 230B in FIG. 9 is a scanning type lamp device and includes a light source 232, a lighting circuit 234, and a scanning optical system 238. The scanning optical system 238 is configured to be able to scan the emitted beam of the light source 232. For example, the scanning optical system 238 may include a motor 239a and a reflector (blade) 239b attached to the rotation axis of the motor 239a. When the motor 239a rotates, the angle formed by the reflecting surface of the reflector 239b and the emitted beam changes, and accordingly, the scanning beam BMSCAN is scanned. A projection optical system 235 of a transmission type or a reflection type may be provided on the optical path of the scanning beam BMSCAN.

[0087] As described above, one aspect of the present invention has been described based on the first embodiment. Next, modification examples related to the first embodiment will be described.

[0088] (Modification Example 1) In the embodiment, the control mode is adaptively switched by the mode controller 224, but the control mode may be manually selectable by the driver. There are individual differences in the characteristics of human eye cells, and there are also individual differences in the light distribution patterns that are felt to be preferable. Therefore, by providing the driver with the freedom to select the control mode, a light distribution pattern appropriate for each driver can be realized.

[0089] (Modification Example 2) The driver may be able to input the parameters referred to when automatically controlling the control mode. Thereby, it is possible to provide appropriate switching of the control mode for each driver.

[0090] (Modification Example 3) In the embodiment, the processing speed limitation of the hardware of the control unit 220 is considered, but not limited thereto. The control unit 220 may also be operable in the region 910 corresponding to the highest resolution and the highest update rate in FIG. 5. Even in this case, the benefits of the present invention can be enjoyed.

[0091] (Modification 4) In the embodiment, the processing mainly focusing on the visibility of the driver's dominant eye has been described, but not limited thereto. In autonomous driving or semi-autonomous driving, object recognition by an in-vehicle camera is important. Therefore, for the in-vehicle camera, the control mode may be adaptively switched so that object recognition is easy.

[0092] (Second Embodiment) FIG. 10 is a block diagram of a vehicle lighting system according to the second embodiment. The vehicle lighting system 100 includes a vehicle lighting device 200, a vehicle ECU (Electronic Control Unit) 102, and a battery 104. The battery voltage VBAT generated by the battery 104 is supplied as a power supply voltage to the vehicle lighting device 200. The vehicle ECU 102 and the vehicle lighting device 200 comply with protocols such as CAN (Controller Area Network) and LIN (Local Interconnect Network) and are communicable. The vehicle ECU 102 controls the lighting and extinguishing of the vehicle lighting device 200 and transmits vehicle information such as vehicle speed and steering to the vehicle lighting device 200. A fail signal or the like can be transmitted from the vehicle lighting device 200 to the vehicle ECU 102.

[0093] The vehicle lamp 200 is configured to be able to dynamically and adaptively control the light distribution pattern PTN based on the image captured by the camera unit 210 (hereinafter referred to as the initial image data IMG1). Note that this initial image data IMG1 corresponds to one frame of data constituting a video. The camera unit 210 may include only one camera, or may include a plurality of cameras with different resolutions and / or frame rates. When the camera unit 210 includes a plurality of cameras, the initial image data IMG1 is understood as a collective term for the output data of the plurality of cameras.

[0094] The light distribution pattern PTN is understood as the two-dimensional illuminance distribution of the irradiation pattern 902 formed by the vehicle lamp 200 on the virtual vertical screen 900 in front of the vehicle. The vehicle lamp 200 includes a control unit (control device) 220 and a light source unit 230 for generating the light distribution pattern PTN. The control unit 220 is also referred to as the lamp ECU.

[0095] The light distribution pattern PTN is divided into a plurality of meshes (individual regions), and the illuminance of the same mesh is constant. The spatial resolution of the light distribution pattern PTN is associated with the fineness (coarseness) of the mesh. Although not limited thereto, for example, the resolution of the light distribution pattern may be any of WUXGA (1920×1200), FHD (1920×1080), FWXGA (1366×768 or 1280×720), SXGA (1280×1024), WXGA (1280×800), WVGA (800×480), VGA (640×480), QVGA (320×240). Alternatively, it may have a coarser resolution, or may have a higher definition resolution equivalent to 4K or 8K.

[0096] Figs. 11(a) to (c) are diagrams for explaining the light distribution pattern PTN. Fig. 11(a) shows an example of the initial image data IMG1, and Figs. 11(b) and (c) are diagrams showing an example of the light distribution pattern PTN corresponding to the initial image data IMG1 in Fig. 11(a). For ease of explanation here, an example will be described in which a specific object target is extracted from the initial image data IMG1, and the mesh at the location where the specific object target exists is shielded from light (i.e., the illuminance is zero). The specific object target is an object that should not be dazzling, and the leading vehicle 904, the oncoming vehicle 906, and the pedestrian 908 are exemplified. In Figs. 11(b) and (c), the shaded areas are hatched. The light distribution pattern PTN in Fig. 11(b) has a higher resolution than the light distribution pattern in Fig. 11(c). The higher the resolution, the more accurately the light shielding conforming to the shape of the specific object target can be realized.

[0097] If the resolution of the light distribution pattern PTN is further increased, it is also possible to shield only the rear window portion of the leading vehicle 904 or only the front window portion of the oncoming vehicle 906, and actively increase the illuminance of the body. Similarly, it is also possible to shield only the face portion of the pedestrian 908 and actively increase the illuminance of the body portion. Thereby, while preventing glare, the visibility of the driver of the host vehicle can be further enhanced.

[0098] Returning to Fig. 10, the specific configuration of the vehicle lamp 200 will be described. The vehicle lamp 200 includes a control unit 220 and a light source unit 230 in addition to the camera unit 210. In Fig. 10, the camera unit 210 is built in the vehicle lamp 200, but it may be provided on the vehicle side.

[0099] The control unit 220 integrally controls the vehicle lamp 200. The control unit 220 includes an image processing unit 221 and other processing units (not shown). The image processing unit 221 can be composed of a digital processor, for example, it may be composed of a combination of a CPU, a microcomputer, and a software program, or it may be composed of an FPGA (Field Programmable Gate Array), an ASIC (Application Specified IC), etc.

[0100] The preprocessing unit 226 receives the initial image data IMG1 captured by the camera unit 210 and generates intermediate image data IMG2 according to the initial image data IMG1.

[0101] Regarding the pixels of the initial image data IMG1 whose pixel values are included in the predetermined range RNG, the effective number of gradations in the intermediate image data IMG2 is smaller than the number of gradations in the initial image data IMG1. In this embodiment, for simplicity, a monochrome image is considered, but a color image may also be used. When generating the intermediate image data IMG2 from the initial image data IMG1, the preprocessing unit 226 may perform a process of reducing the resolution.

[0102] The light distribution pattern generator 222 generates illumination control data that defines the illumination distribution (light distribution pattern PTN) of the vehicle lamp 200 based on the intermediate image data IMG2. The method for generating the light distribution pattern PTN based on the intermediate image data IMG2 is not particularly limited, but the illumination of the individual regions (meshes) included in the light distribution pattern PTN is based on the value (pixel value) of the corresponding pixel of the intermediate image data IMG2.

[0103] The light source unit 230 receives the illumination control data that instructs the light distribution pattern PTN from the light distribution pattern generator 222 and is configured to be able to form an illumination distribution corresponding to the light distribution pattern PTN in front of the vehicle. The configuration of the light source unit 230 is not particularly limited, and for example, it may include a semiconductor light source such as an LD (laser diode) or an LED (light emitting diode), and a lighting circuit that drives the semiconductor light source to turn it on. The light source unit 230 may include a matrix-type pattern forming device such as a DMD (Digital Mirror Device) or a liquid crystal device for forming an illumination distribution corresponding to the light distribution pattern PTN.

[0104] The above is the configuration of the vehicle lamp 200. Next, its operation will be described. FIG. 12 is a diagram for explaining the signal processing in the vehicle lamp 200 of FIG. 10. In FIG. 12, for simplicity of explanation, the image data is shown simplified one-dimensionally. A predetermined range RNG, which is the target of the gradation number reduction process, is hatched. In this example, the 12 gradations from pixel value 0 to 11 are the predetermined range RNG. Referring to the intermediate image data IMG2, the effective gradation number of the predetermined range RNG is 3 gradations.

[0105] The original initial image data IMG1 randomly includes noise N. Among this noise, the noise included in the predetermined range RNG can be removed by reducing the gradation number.

[0106] Note that the process of reducing the gradation number may be a process of rounding (rounding up or down) the lower several bits. In the example of FIG. 12, the rounding process is performed by setting the lower 2 bits to zero. How many lower bits to round, in other words, how much to reduce the gradation number, may be defined in consideration of the amplitude of the noise N and the influence that the noise N has on the final light distribution pattern. Therefore, the image processing unit 221 may change the degree of reducing the gradation number according to the amplitude of the noise (in other words, the S / N ratio) included in the initial image data IMG1.

[0107] The above is the operation of the vehicle lamp 200. According to this vehicle lamp 200, among the pixels included in the initial image data IMG1, regarding the pixels where the influence of noise becomes significant in the final illuminance distribution, by reducing the effective gradation number, a light distribution with reduced influence of noise derived from the camera can be formed. Also, by reducing the gradation number only within the predetermined range RNG with hatching, for the pixels in the other ranges where the noise is less noticeable, the original high gradation number can be maintained.

[0108] The present invention extends to various devices, circuits, and methods that can be understood from the block diagrams and circuit diagrams of FIG. 10 or derived from the above description, and is not limited to a specific configuration. Hereinafter, in order to assist in understanding the essence of the invention and the circuit operation, and to clarify them, rather than narrowing the scope of the present invention, more specific configuration examples and modification examples will be described.

[0109] The relationship between the image processing in the light distribution pattern generator 222 and the generation of intermediate image data will be described in detail with reference to several examples.

[0110] (First Embodiment) FIG. 13 is a diagram for explaining the image processing according to the first embodiment. Most simply, the light distribution pattern generator 222 may detect a specific target based on the initial image data IMG1 and perform control to block (or reduce the intensity of) the detected target portion. For example, examples of objects to be blocked include light emitters such as oncoming vehicles, preceding vehicles, street lights, and electronic billboards. For oncoming vehicles and preceding vehicles, blocking is required from the perspective of glare suppression, and for light emitters such as street lights and electronic billboards, it is meaningless to irradiate the headlight in the first place.

[0111] Therefore, the image processing unit 221 may determine that the range where the probability of the presence of a light emitter is high for the pixels in the initial image data IMG1 whose pixel values exceed a predetermined threshold TH, and perform control to block (or reduce the intensity of) them.

[0112] When performing such control, when the pixel value of the initial image data IMG1 is near the threshold TH, the pixel value with noise superimposed thereon will cross the threshold TH, and a blocking area will be formed according to random noise. This state is shown in the light distribution pattern in the third row from the top in FIG. 13.

[0113] Therefore, by defining the predetermined range RNG and the number of gray levels of the intermediate image data within the predetermined range RNG in association with the amplitude of the noise and the threshold TH, the influence of the noise can be reduced. This state is shown in the light distribution pattern in the fourth row from the top in FIG. 13.

[0114] (Second Embodiment) The light distribution pattern generator 222 may perform processing to reduce or increase the contrast of the forward field of view of the vehicle based on the initial image data IMG1. The contrast here is understood as the ratio of the brightness between the dark part and the bright part.

[0115] More specifically, to reduce the contrast, it is only necessary to increase the illuminance of the dark part (the part with low reflectance), and conversely, to reduce the illuminance of the bright part. In other words, according to the brightness of the irradiated object, the brighter the object, the lower the illuminance, and the darker the object, the higher the illuminance is set. That is, for each individual area of the light distribution pattern, the lower the pixel value of the corresponding pixel in the initial image data IMG1, the higher the illuminance is set, and the higher the pixel value, the lower the illuminance is set. In this specification, this processing is referred to as low-contrast control.

[0116] FIG. 14(a) is a diagram for explaining the low-contrast control. As a result, the brightness difference between the dark part and the bright part becomes smaller, and in particular, the visibility of the dark part can be improved. In addition to the solid line of the straight line sloping down to the right, stepped discrete control sloping down to the right as shown by the broken line may be performed, or control along a curve as shown by the dashed-dotted line may be performed. Alternatively, the illuminance of the mesh may be adjusted by feedback control so that the pixel value of each mesh approaches a predetermined target value.

[0117] On the contrary, to enhance the contrast, the illuminance of the dark part (the part with low reflectance) should be lowered, and conversely, the illuminance of the bright part should be increased. In other words, according to the brightness of the irradiated object, the higher the brightness of the object, the higher the illuminance is set, and the darker the object, the lower the illuminance is set. That is, for each individual area of the light distribution pattern, the lower the pixel value of the corresponding pixel in the initial image data IMG1, the lower the illuminance is set, and the higher the pixel value, the higher the illuminance is set. In this specification, this process is called high-contrast processing. FIG. 14(b) is a diagram for explaining high-contrast control. In high-contrast control, thereby, the brightness difference between the dark part and the bright part becomes larger. A visual field with a large brightness difference has the advantage that the human eye can easily identify the position and shape of an object instantaneously. In addition to the control based on the rising straight line on the right shoulder, as shown by the broken line, a stepped control rising on the right shoulder may be performed, or a control along a curve as shown by the dashed-dotted line may be performed.

[0118] FIG. 15 is a diagram for explaining the image processing according to the second embodiment. Here, a case where the low-contrast control shown in FIG. 14(a) is realized by tone inversion will be described. For example, when the maximum tone is MAX and the pixel value of a certain pixel is X, the inverted tone Y is given by the following formula. Y = MAX - X

[0119] When directly performing tone inversion on the initial image data IMG1, the low tones move to the high-tone region of the irradiation distribution. Therefore, the noise included in the low tones of the initial image data IMG1 will be included in the region with high illuminance and become conspicuous (the upper right in FIG. 15 without rounding). Specifically, within the bright region of the light distribution pattern, there will be spots that become dark due to noise, which may reduce the visibility.

[0120] Therefore, instead of directly performing tone inversion on the initial image data IMG1, the pixel values are rounded in the low-tone range (0 to A) in the initial image data IMG1 to generate intermediate image data IMG2, and by performing tone inversion on the intermediate image data IMG2, the noise in the region with high illuminance can be reduced (with rounding in the lower right of FIG. 15).

[0121] In summary, the following technical idea can be derived. That is, the image processing unit 221 may define a predetermined range RNG so that the noise included in the illuminance range where the noise is prominent in the illuminance distribution is reduced. When the pixel value is X and the corresponding illuminance is Y, it is assumed that the relationship between them is represented by an arbitrary function. Y = f(X) At this time, when the illuminance distribution includes noise, the upper limit value of the illuminance range where the noise is prominent is YMAX, and the lower limit value is YMIN. The pixel values XMAX and XMIN corresponding to YMAX and YMIN are given by the following equations. XMAX = f-1(YMAX) XMIN = f-1(YMIN) f-1 is the inverse function of the function f. Therefore, the upper and lower limits of the predetermined range may be defined by XMAX and XMIN. Note that there may be a case where XMAX > XMIN, or there may be a case where XMAX < XMIN.

[0122] (Third Embodiment) Another example of contrast control will be described. FIGS. 16(a) and 16(b) are diagrams for explaining the contrast control according to the third embodiment. FIG. 16(a) shows the relationship between the reflectance (%) of an object and the brightness. When irradiated with a constant illuminance independent of the subject, as shown in (i), the brightness of the object is proportional to the reflectance. A high-contrast image in the so-called field of image processing has a tone distribution as shown in (ii), and a low-contrast image may have a tone distribution as shown in (iii).

[0123] FIG. 16(b) shows the relationship between the reflectance of the object and the illuminance (normalized relative value). The reflectance of the object may be read as the pixel value. In both cases of high-contrast control or low-contrast control, in the region where the reflectance is low, the slope of the reflectance-illuminance characteristic is large. This means that the amplitude of the noise in the low-tone region of the pixel value is amplified. Therefore, when performing the contrast control as shown in FIG. 16(a), it is advisable to perform rounding processing on the low-tone region as the predetermined range RNG to generate the intermediate image data IMG2.

[0124] In summary, the following technical idea can be derived. That is, when the image processing unit 221 converts (maps) from pixel values to illuminance, if there is a gradation range in which the amplitude of noise is amplified, the gradation range may be defined by a predetermined range RNG.

[0125] When the pixel value is X and the corresponding illuminance is Y, assume that the relationship between them is represented by an arbitrary function. Y = f(X) The range where the noise amplification rate is high corresponds to the range where the slope of the function f(X) is large and the absolute value of the derivative function f'(X) is large. Therefore, when the pixel value that gives the maximum value of the slope |f'(X)| is XMAX, the predetermined range RNG may be defined so as to include the pixel value XMAX.

[0126] As described above, one aspect of the present invention has been described based on the second embodiment. Subsequently, a modification related to the second embodiment will be described.

[0127] (Modification 1) In the embodiment, noise countermeasures mainly from the driver's perspective have been described, but it is not limited to that. In autonomous driving or semi-autonomous driving, since object recognition by an in-vehicle camera is important, a predetermined range RNG for performing rounding processing may be determined so that noise that is not preferable for the in-vehicle camera is reduced.

[0128] (Modification 2) In the embodiment, the case where visibility deteriorates when a spot that becomes dark due to noise is included in the bright area of the light distribution pattern has been described, but it is not limited to that. The human eye is more sensitive to luminance changes in dark areas than to luminance changes in bright areas. From this perspective, when a spot that becomes bright due to noise is included in the dark area of the light distribution pattern, it may be unpleasant in some cases. Therefore, in this case, the predetermined range RNG may be defined so that noise included in the dark gradation of illuminance is reduced.

[0129] (Modification 3) In the embodiment, bit truncation (rounding up) has been described as a rounding process for reducing the number of gradation levels, but this is not the only case. For pixel values included in a predetermined range RNG, the substantial number of gradation levels may be reduced by multiplying by a coefficient smaller than 1. This process is effective when the predetermined range RNG is a low gradation region.

[0130] Based on the embodiment, the present invention has been described using specific terms. However, the embodiment merely shows one aspect of the principle and application of the present invention. In the embodiment, many modifications and arrangement changes are permitted without departing from the idea of the present invention defined in the claims.

Explanation of Signs

[0131] 100... Vehicle lighting system, 102... Vehicle ECU, 104... Battery, 200... Vehicle lighting fixture, 210... Camera unit, 220... Control unit, 221... Image processing unit, 222... Light distribution pattern generator, 224... Mode controller, 226... Preprocessing unit, 230... Light source unit, 232... Light source, 234... Lighting circuit, 236... Patterning device, 238... Scanning optical system, 240... Illuminance sensor, PTN... Light distribution pattern, IMG... Camera image, 900... Virtual vertical screen, 902... Irradiation pattern, 904... Preceding vehicle, 906... Oncoming vehicle, 908... Pedestrian.

Industrial Applicability

[0132] The present invention can be used for vehicle lighting fixtures.

Claims

1. A control device for a vehicle lamp, comprising an image processing unit that receives initial image data captured by a camera, generates intermediate image data according to the initial image data, and generates illuminance control data that defines an illuminance distribution of the vehicle lamp based on the intermediate image data, wherein, with respect to pixels whose pixel values are included in a predetermined range among the pixels of the initial image data, an effective number of gradations in the intermediate image data is smaller than the number of gradations in the initial image data, the image processing unit performs a process of converting the pixel value into illuminance, and the predetermined range is defined such that noise included in an illuminance range where the noise is prominent in the illuminance distribution is reduced.

2. A control device for a vehicle lamp, comprising an image processing unit that receives initial image data captured by a camera, generates intermediate image data according to the initial image data, and generates illuminance control data that defines an illuminance distribution of the vehicle lamp based on the intermediate image data, wherein, with respect to pixels whose pixel values are included in a predetermined range among the pixels of the initial image data, an effective number of gradations in the intermediate image data is smaller than the number of gradations in the initial image data, the image processing unit performs a process of converting the pixel value into illuminance, and when a relationship of Y = f(X) holds using a function f(X) with the pixel value as X and the illuminance as Y, the predetermined range is defined to include a pixel value XMAX that gives a maximum value of an absolute value |f'(X)| of a derivative function f'(X) of the function f(X).

3. The process of generating the illuminance control data includes a process of comparing a pixel value of the intermediate image data or image data obtained therefrom with a threshold value, and the predetermined range is defined to include the threshold value. The control device for a vehicle lamp according to Claim 1 or 2.

4. A control device for a vehicle lamp, comprising an image processing unit that receives initial image data captured by a camera, generates intermediate image data according to the initial image data, and generates illuminance control data that defines an illuminance distribution of the vehicle lamp based on the intermediate image data, wherein, with respect to pixels whose pixel values are included in a predetermined range among the pixels of the initial image data, an effective number of gradations in the intermediate image data is smaller than the number of gradations in the initial image data, the process of generating the illuminance control data includes a process of comparing a pixel value of the intermediate image data or image data obtained therefrom with a threshold value, A control device for a vehicle lamp, wherein the predetermined range is defined to include the threshold value. **Claim 5** The process of generating the illuminance control data includes a process of changing the contrast of the intermediate image data or the image data obtained therefrom. The control device for a vehicle lamp according to any one of claims 1 to 4. **Claim 6** The process of generating the illuminance control data includes a process of inverting the pixel values of the intermediate image data or the image data obtained therefrom. The control device for a vehicle lamp according to any one of claims 1 to 5. **Claim 7** An image processing unit that receives initial image data captured by a camera, generates intermediate image data corresponding to the initial image data, and generates illuminance control data that defines the illuminance distribution of a vehicle lamp based on the intermediate image data. With respect to the pixels among the pixels of the initial image data whose pixel values are included in a predetermined range on the low gradation side including 0, the intermediate image data is generated such that the effective number of gradations in the intermediate image data is smaller than the number of gradations in the initial image data. The process of generating the illuminance control data includes a process of inverting the pixel values of the intermediate image data or the image data obtained therefrom. The control device for a vehicle lamp. **Claim 8** The process of generating the intermediate image data includes a process of rounding the lower N bits (N is an integer of N≥1) of the pixel values of the initial image data. The control device for a vehicle lamp according to any one of claims 1 to 7. **Claim 9** The process of generating the intermediate image data includes a process of multiplying the pixel values of the initial image data by a positive coefficient smaller than 1. The control device for a vehicle lamp according to any one of claims 1 to 7. **Claim 10** The image processing unit changes the degree of reducing the number of gradations according to the amplitude of the noise included in the initial image data. The control device for a vehicle lamp according to any one of claims 1 to 9. **Claim 11** An image processing unit that receives initial image data captured by a camera, generates intermediate image data corresponding to the initial image data, and generates illuminance control data that defines the illuminance distribution of a vehicle lamp based on the intermediate image data. With respect to the pixels among the pixels of the initial image data whose pixel values are included in a predetermined range, the effective number of gradations in the intermediate image data is smaller than the number of gradations in the initial image data. The image processing unit is characterized in that the degree of reducing the number of gradations is changed according to the amplitude of noise included in the initial image data, in a control device for a vehicle lamp.

12. A vehicle lamp characterized by comprising the control device according to any one of Claims 1 to 11.

13. A control method for a vehicle lamp, comprising: a step of generating intermediate image data according to initial image data captured by a camera; a step of generating illuminance control data defining an illuminance distribution of the vehicle lamp based on the intermediate image data; and with respect to pixels whose pixel values are included in a predetermined range among the pixels of the initial image data, the effective number of gradations in the intermediate image data is smaller than the number of gradations in the initial image data, the step of generating the illuminance control data includes a step of performing a process of converting the pixel value into illuminance, wherein the predetermined range is defined such that noise included in an illuminance range where noise is prominent in the illuminance distribution is reduced, in a control method for a vehicle lamp.

Images