WO2018040756A1 - 一种车身颜色识别的方法及装置 - Google Patents
一种车身颜色识别的方法及装置 Download PDFInfo
- Publication number
- WO2018040756A1 WO2018040756A1 PCT/CN2017/092651 CN2017092651W WO2018040756A1 WO 2018040756 A1 WO2018040756 A1 WO 2018040756A1 CN 2017092651 W CN2017092651 W CN 2017092651W WO 2018040756 A1 WO2018040756 A1 WO 2018040756A1
- Authority
- WO
- WIPO (PCT)
- Prior art keywords
- color
- channel
- image
- color channel
- distribution
- Prior art date
Links
- 238000000034 method Methods 0.000 title claims abstract description 35
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 17
- 238000009826 distribution Methods 0.000 claims description 91
- 238000012706 support-vector machine Methods 0.000 claims description 42
- 238000004364 calculation method Methods 0.000 claims description 25
- 238000012545 processing Methods 0.000 claims description 18
- 238000005286 illumination Methods 0.000 abstract description 4
- 238000013461 design Methods 0.000 description 12
- 238000010586 diagram Methods 0.000 description 8
- 238000004590 computer program Methods 0.000 description 6
- 238000012986 modification Methods 0.000 description 4
- 230000004048 modification Effects 0.000 description 4
- 230000006870 function Effects 0.000 description 3
- 238000012544 monitoring process Methods 0.000 description 3
- 238000003860 storage Methods 0.000 description 3
- 239000003086 colorant Substances 0.000 description 2
- 238000010276 construction Methods 0.000 description 1
- 230000001186 cumulative effect Effects 0.000 description 1
- 238000011835 investigation Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012806 monitoring device Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/56—Extraction of image or video features relating to colour
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2411—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on the proximity to a decision surface, e.g. support vector machines
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/50—Context or environment of the image
- G06V20/52—Surveillance or monitoring of activities, e.g. for recognising suspicious objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V2201/00—Indexing scheme relating to image or video recognition or understanding
- G06V2201/08—Detecting or categorising vehicles
Definitions
- the invention relates to the field of computer vision, and in particular to a method and a device for color recognition of a vehicle body.
- body color recognition is particularly important in the intelligent monitoring field of safe city construction.
- vehicle license plate information is difficult to identify and the vehicle information is difficult to identify, the body color information is the most obvious feature. It plays a decisive role in case investigation, criminal tracking, and deck identification.
- the body color is metallic, when the color of the car body is automatically recognized in the intelligent monitoring scene, the influence of the illumination is large, and the recognized color accuracy of the vehicle body is low.
- an embodiment of the present invention provides a method for color recognition of a vehicle body, the method comprising: acquiring a first image including a target vehicle, and setting a first image in a Hue Saturation Value (HSV) color space. Performing a histogram equalization operation on the luminance V channel to determine a second image after the histogram equalization operation; dividing the second image into a plurality of different regions by sliding the setting window in a set order Determining a first color feature corresponding to each of the plurality of different regions included in the second image; and determining, according to the determined different first color features, the first color corresponding to the plurality of different regions determined The feature determines the body color of the target vehicle according to a preset algorithm.
- HSV Hue Saturation Value
- the second image includes a plurality of color spaces
- the first color feature includes at least one of a distribution of a target pixel point of the first color channel and a color moment of the first color channel.
- the second image includes a plurality of color channels, the first color channel is one of the plurality of color channels, and the distribution of the first color channel target pixel points describes that the pixel values in the first color channel are
- N is a positive integer greater than or equal to 1. In the embodiment of the present invention, N is a positive integer of 4 or more.
- the first image is subjected to a histogram equalization operation on the V channel, the influence of the illumination on the color of the vehicle body is removed, and the color of the vehicle body is recognized according to the determined different first color features, thereby improving the color recognition of the vehicle body. Accuracy.
- the first color feature is a distribution of target pixel points of the first color channel
- the distribution of the target pixel points of the first color channel corresponding to the plurality of different regions to be determined is determined.
- the linear support vector machine determining a body color of the target vehicle according to a first calculation model corresponding to a distribution of the first color channel target pixel points.
- the embodiment of the invention identifies the color of the vehicle body according to the determined distribution of the target pixel points of the first color channel, thereby improving the accuracy of the color recognition of the vehicle body.
- the linear support vector machine determining a body color of the target vehicle according to a second calculation model corresponding to a color moment of the first color channel, wherein the color moment comprises a first moment, two Moment and third moment.
- the embodiment of the invention identifies the color of the vehicle body according to the determined color moment of the first color channel, thereby improving the accuracy of the color recognition of the vehicle body.
- the first color feature is a distribution of a first color channel target pixel point and a color moment of the first color channel
- the determined plurality of different regions are corresponding to the a distribution of the first color channel target pixel point and a color moment of the first color channel are input to a linear support vector machine, and the linear support vector machine is configured according to a distribution of the target pixel point with the first color channel
- the third calculation model corresponding to the color moment of the first color channel determines the body color of the target vehicle.
- the embodiment of the invention identifies the color of the vehicle body according to the determined distribution of the target pixel points of the first color channel and the color moment of the first color channel, thereby improving the accuracy of the color recognition of the vehicle body.
- the first image before performing the histogram equalization operation on the V channel of the HSV color space, the first image further includes:
- the first image is adjusted to a set size.
- Determining the first image of the set size improves the accuracy of the data processed by the linear support vector machine.
- adjacent ones of the plurality of regions of the second image partially overlap.
- each pixel in the second image is ensured to be statistically improved, and the accuracy of the color recognition of the vehicle body is improved.
- an embodiment of the present invention provides a method for color recognition of a vehicle body, the method comprising: acquiring a first image including a target vehicle, and performing a first image on a luminance V channel of a hue saturation luminance HSV color space.
- a picture equalization operation determining a second image after the histogram equalization operation; dividing the second image into a plurality of first different areas by sliding the setting window in a set order, and determining the Determining a first color feature corresponding to each of the plurality of first different regions included in the second image; selecting at least one region of the partial image information of the target vehicle in the first image according to a preset rule, Performing a histogram equalization operation on the V channel of the HSV color space to determine at least one third image after the histogram equalization operation; sliding the setting window in a set order to divide the third image Determining a second color feature corresponding to the plurality of second different regions for the plurality of second different regions, according to the determined plurality of first different region pairs Wherein the first color and the second plurality of regions of different characteristic corresponding to a second color, according to a preset algorithm, it is determined that the target vehicle body color.
- the second image includes a plurality of color spaces
- the first color feature includes at least one of a distribution of a target pixel point of the first color channel and a color moment of the first color channel.
- the second image includes a plurality of color channels, the first color channel is one of the plurality of color channels, and the distribution of the first color channel target pixel points describes that the pixel values in the first color channel are Each set a number of pixels within a threshold range, the set threshold range is N, N is a positive integer greater than or equal to 1; and the second color feature includes a distribution of target pixel points of the second color channel At least one of a situation and a color moment of the second color channel, the third image comprising a plurality of color channels, the second color channel being the plurality of One of the color channels, the distribution of the second color channel target pixel points describes the number of pixel points in the second color channel whose pixel values are within each set threshold range, the set threshold range is M M is a positive integer greater than or equal
- the first image and the third image are subjected to a histogram equalization operation on the V channel, thereby removing the influence of the illumination on the color of the vehicle body, and according to the determined different first color features and different second color feature pairs.
- the body color is identified to improve the accuracy of body color recognition.
- the first color feature is a distribution of target pixel points of a first color channel
- the second color feature is a distribution of target pixel points of a second color channel
- the determined a distribution of the first color channel target pixel points corresponding to the plurality of first different regions and a distribution of the second color channel target pixel points corresponding to the plurality of second different regions
- the linear support vector machine determines the body color of the target vehicle according to a first calculation model corresponding to the distribution of the first color channel target pixel point and the distribution of the second color channel target pixel point.
- the embodiment of the invention identifies the color of the vehicle body according to the determined distribution of the target pixel points of the first color channel and the distribution of the target pixel points of the second color channel, thereby improving the accuracy of the color recognition of the vehicle body.
- the first color feature is the color moment of the first color channel and the second color feature is the color moment of the second color channel
- multiple a color moment of the first color channel corresponding to the first different area and a color moment of the first color channel corresponding to the plurality of second different areas are input to a linear support vector machine, and the linear support vector machine is based on a second calculation model corresponding to a color moment of the first color channel and a color moment of the second color channel, determining a body color of the target vehicle, wherein the color moment comprises a first moment, a second moment, and Third moment.
- the embodiment of the invention identifies the color of the vehicle body according to the determined color moment of the first color channel and the color moment of the second color channel, thereby improving the accuracy of the body color recognition.
- the first color feature is a distribution of a first color channel target pixel point and a color moment of the first color channel
- the second color feature is a second color channel target pixel a distribution of the points and a color moment of the second color channel
- the distribution of the second color channel target pixel points corresponding to the plurality of second different regions and the color moment of the second color channel are input to a linear support vector machine, and the linear support vector machine is configured according to a distribution pattern of a color channel target pixel point and a color moment of the first color channel, and a distribution pattern of the second color channel target pixel point and a third calculation model corresponding to the color moment of the second color channel
- the body color of the target vehicle is determined.
- the embodiment of the present invention identifies the vehicle body color according to the determined distribution of the first color channel target pixel point and the color moment of the first color channel, and the distribution of the second color channel target pixel point and The color moment of the second color channel improves the accuracy of the body color recognition.
- adjacent ones of the plurality of second different regions of the third image partially overlap.
- each pixel in the third image is ensured to be statistically improved, and the color recognition of the vehicle body is improved. Accuracy.
- an embodiment of the present invention provides a device for color recognition of a vehicle body, the device comprising:
- a processing module configured to perform a histogram equalization operation on the luminance V channel of the hue saturation luminance HSV color space to obtain a second image, where the first image includes image information of the target vehicle;
- a dividing module configured to slide the setting window in a set order, and divide the second image into a plurality of different regions
- a determining module configured to determine a first color feature corresponding to the plurality of different regions, the first color feature including a distribution of a first color channel target pixel point and at least a color moment of the first color channel
- the second image includes a plurality of color channels, the first color channel is one of the plurality of color channels, and the distribution of the first color channel target pixel points is described in the first color channel.
- a number of pixel points whose pixel value is within each set threshold range, the set threshold range is N, and N is a positive integer greater than or equal to 1;
- an identifying module configured to determine, according to a preset algorithm, a body color of the target vehicle according to the determined first color feature corresponding to the plurality of different regions.
- the identifying module is configured to:
- the linear support vector machine inputting, to the linear support vector machine, the distribution of the first color channel target pixel points corresponding to the plurality of different regions, the linear support vector machine according to the distribution of the target pixel points with the first color channel Corresponding the first calculation model determines the body color of the target vehicle.
- the identification module is configured to:
- the linear support vector machine Inputting the determined color moments of the first color channel corresponding to the plurality of different regions to the linear support vector machine, the linear support vector machine according to the second calculation model corresponding to the color moment of the first color channel Determining the body color of the target vehicle.
- the identification module is configured to:
- determining, by the determined plurality of different regions, a distribution of the first color channel target pixel points and a color moment of the first color channel to a linear support vector machine, wherein the linear support vector machine is configured according to A third calculation model corresponding to a distribution of the first color channel target pixel point and a color moment of the first color channel determines a body color of the target vehicle.
- an embodiment of the present invention provides a device for color recognition of a vehicle body, the device comprising:
- a first processing module configured to perform a histogram equalization operation on the luminance V channel of the hue saturation luminance HSV color space to obtain a second image, where the first image includes image information of the target vehicle;
- a first dividing module configured to slide the setting window according to the setting order, and divide the second image into a plurality of first different regions
- a first determining module configured to determine a first color feature corresponding to the plurality of first different regions, where the first color feature includes a distribution of a first color channel target pixel point and a color of the first color channel At least one of the moments, the second image includes a plurality of color channels, the first color channel is one of the plurality of color channels, and the distribution of the target pixel points of the first color channel describes a number of pixel points in a color channel in each set threshold value range, the set threshold range is N, and N is a positive integer greater than or equal to 1;
- An acquiring module configured to select at least one area in the first image according to a preset rule, where the at least one area includes partial image information of the target vehicle;
- a second processing module configured to perform at least one third image by performing a histogram equalization operation on the V channel of the HSV color space;
- a second dividing module configured to slide the setting window according to the setting order, and divide the third image into a plurality of second different regions
- a second determining module configured to determine a second color feature corresponding to the plurality of second different regions, where the second color feature includes a distribution of the second color channel target pixel point and a color of the second color channel At least one of the moments, the third image includes a plurality of color channels, the second color channel is one of the plurality of color channels, and the distribution of the second color channel target pixel points describes The number of pixel points in the two color channels in which the pixel value is within each set threshold range, the set threshold range is M, and M is a positive integer greater than or equal to 1;
- An identification module configured to determine, according to a preset algorithm, the target vehicle according to the determined first color feature corresponding to the plurality of first different regions and the second color feature corresponding to the plurality of second different regions Body color.
- an embodiment of the present invention provides a vehicle body color recognition device, including a processor, and a memory connected to the processor, wherein:
- a memory for storing program code executed by the processor
- a processor configured to execute the program code stored by the memory, performs the following process:
- the first color feature corresponding to the plurality of different regions, the first color feature comprising at least one of a distribution of a first color channel target pixel point and a color moment of the first color channel, the first
- the second image includes a plurality of color channels, the first color channel is one of the plurality of color channels, and the distribution of the first color channel target pixel points describes the pixel values in the first color channel in each Setting a number of pixels in the threshold range, the set threshold range is N, and N is a positive integer greater than or equal to 1;
- the solution provided by the invention improves the accuracy of body color recognition.
- FIG. 1 is a schematic flow chart of a method for identifying a color of a vehicle body according to an embodiment of the present invention
- FIG. 2 is a schematic flowchart diagram of still another body color recognition method according to an embodiment of the present invention.
- FIG. 3 is a schematic flowchart diagram of another body color recognition method according to an embodiment of the present invention.
- FIG. 4 is a schematic structural diagram of a vehicle body color recognition device according to an embodiment of the present invention.
- FIG. 5 is a schematic structural diagram of still another body color recognition device according to an embodiment of the present invention.
- FIG. 6 is a schematic structural diagram of hardware of a vehicle body color recognition device according to an embodiment of the present invention.
- An embodiment of the present invention provides a method for color recognition of a vehicle body. As shown in FIG. 1, the method includes the following process:
- the processor performs a histogram equalization operation on the luminance V channel of the hue saturation luminance HSV color space to obtain a second image, as shown in FIG. 2, wherein the HSV color space is a hexagonal cone.
- the volume model which is described in terms of hue saturation brightness, is often used in image processing.
- the method further includes the processor adjusting the acquired first image including the target vehicle information to a set size.
- the value of each channel in the red green blue (RGB) color space in the first image is determined, and the value of each channel in the RGB color space is converted into the value of each channel in the HSV color space, specifically as follows:
- r' is the value of the R channel in the RGB color space
- g' is the value of the G channel in the RGB color space
- b' is the value of the B channel in the RGB color space
- h' is the value of the H channel in the HSV color space
- s' is the value of the S channel in the HSV color space
- v' is the value of the V channel in the HSV color space
- max is the maximum of the three channels R, G, and B in the RGB color space
- min is the R in the RGB color space.
- G, B The minimum of the three channels.
- a histogram of the number of pixels of the same v value in the V channel is counted, where v is [0, 255], A total of 256 levels;
- H are the height and width of the second image, respectively, and H ⁇ W represents the total number of pixels in the second image.
- the equalized HSV color space is converted into a new RGB color space, as follows:
- r is the value of the R channel in the new RGB color space
- g is the value of the G channel in the RGB color space
- b is the value of the B channel in the new RGB color space.
- the RGB color space is a three-dimensional cube model, and the model is described in terms of physical three primary colors, and is often applied to image processing.
- the processor slides the setting window according to the setting order, and divides the second image into a plurality of different regions.
- the second image may be divided into multiple regions by other methods, which is not limited by the present invention.
- adjacent ones of the plurality of regions of the second image partially overlap.
- the first color feature includes at least one of a distribution of a first color channel target pixel point and a color moment of the first color channel
- the second image includes a plurality of color channels.
- the first color channel is one of the plurality of color channels
- the distribution of the first color channel target pixel points describes pixel points in the first color channel that have pixel values within each set threshold range.
- Number, the setting The threshold range is N, and N is a positive integer greater than or equal to 1.
- each image may be represented by a plurality of different color spaces, each color space corresponding to multiple color channels, that is, each image has multiple color channels, and the color channel is used to save image color information. Channel.
- the division of the threshold range is not limited, and is specifically determined by actual needs.
- 0 to 255 are divided into four consecutive threshold ranges, and the four consecutive threshold ranges are 0 to 63, 64 to 127, 128 to 190, and 191 to 255, respectively.
- the color moments of any color channel are divided into first moment, second moment and third moment, and the specific calculation is as follows:
- N represents the total number of pixels in the corresponding region.
- the processor inputs the determined different first color feature processor to the pre-trained linear support vector machine to determine the body color of the target vehicle.
- the calculation model corresponding to the linear support vector machine is different.
- the first image is subjected to a histogram equalization operation on the V channel of the HSV color space to obtain a second image, where the first image includes image information of the target vehicle; and the setting window is swept in the set order.
- the processor adopts a class in the first image adjusted to the set size.
- the manner of the pyramid determines an area including the target vehicle local information, and performs a histogram equalization operation on the V channel of the HSV color space to obtain a third image, and the specific equalization
- the operation process is the same as the above step S11, and will not be described herein.
- the threshold is The range is at least four, that is, 0 to 255, and is divided into four threshold ranges.
- the division of the threshold range is not limited, and is specifically determined by actual needs.
- the calculation method of the color moment in the second color feature is consistent with the calculation method of the color moment in the first color feature, and will not be described herein.
- the determined different first color features and different second color features are input to a pre-trained linear support vector machine to determine the body color of the target vehicle.
- the first color feature and the second color feature are used to determine the body color of the target vehicle, thereby improving the accuracy of the body color recognition.
- the second image in addition to determining a region including the target vehicle local information by using a pyramid-like manner, other pyramid forms may be adopted, for example, dividing the second image into four regions including target vehicle information, and then After the histogram equalization processing is performed on each area, the second color feature corresponding to the area is extracted, and the second image may be divided in other manners in the embodiment of the present invention, which is not limited by the present invention.
- an apparatus for color recognition of a vehicle body includes:
- the processing module 41 is configured to perform a histogram equalization operation on the luminance V channel of the hue saturation luminance HSV color space to obtain a second image, where the first image includes image information of the target vehicle;
- a dividing module 42 configured to slide the setting window according to the setting order, and divide the second image into a plurality of different regions
- a determining module 43 configured to determine a first color feature corresponding to the plurality of different regions, where the first color feature includes a distribution of a first color channel target pixel point and a color moment of the first color channel At least one, the second image includes a plurality of color channels, the first color channel is one of the plurality of color channels, and the distribution of the first color channel target pixel points describes the first color channel
- the number of pixels in which the pixel value is within each set threshold range, the set threshold range is N, and N is a positive integer greater than or equal to 1;
- the identifying module 44 is configured to follow the first color feature corresponding to the determined plurality of different regions.
- An algorithm is provided to determine the body color of the target vehicle.
- the first image is subjected to a histogram equalization operation on the V channel of the HSV color space to obtain a second image, where the first image includes image information of the target vehicle; and the setting window is swept in the set order.
- the identifying module is configured to: determine the first color channel target pixel corresponding to the plurality of different regions The distribution is input to a linear support vector machine that determines the body color of the target vehicle based on a first calculation model corresponding to the distribution of the first color channel target pixel points.
- the identifying module is configured to: determine a color moment of the first color channel corresponding to the plurality of different regions, Input to a linear support vector machine that determines a body color of the target vehicle based on a second calculation model corresponding to a color moment of the first color channel.
- the identifying module is configured to: determine the determined plurality of different regions
- the distribution of the first color channel target pixel point and the color moment of the first color channel are input to a linear support vector machine, and the linear support vector machine is distributed according to the target pixel point of the first color channel
- a third calculation model corresponding to the color moment of the first color channel determines a body color of the target vehicle.
- an apparatus for color recognition of a vehicle body includes:
- the first processing module 51 is configured to perform a histogram equalization operation on the luminance V channel of the hue saturation luminance HSV color space to obtain a second image, where the first image includes image information of the target vehicle. ;
- the first dividing module 52 is configured to slide the setting window according to the setting order, and divide the second image into a plurality of first different regions;
- a first determining module 53 configured to determine a first color feature corresponding to the plurality of first different regions, where the first color feature includes a distribution of a first color channel target pixel point and the first color channel At least one of the color moments, the second image includes a plurality of color channels, the first color channel is one of the plurality of color channels, and the distribution of the target pixel points of the first color channel describes a number of pixels in the first color channel whose pixel value is within each set threshold range, the set threshold range is N, and N is a positive integer greater than or equal to 1;
- the obtaining module 54 is configured to select at least one area in the first image according to a preset rule, where the at least one area includes partial image information of the target vehicle;
- a second processing module 55 configured to perform the histogram equalization operation on the V channel of the HSV color space to obtain at least one third image
- a second dividing module 56 configured to slide the setting window according to the setting order, and divide the third image into a plurality of second different regions;
- a second determining module 57 configured to determine a second color feature corresponding to the plurality of second different regions,
- the second color feature includes at least one of a distribution of the second color channel target pixel point and a color moment of the second color channel
- the third image includes a plurality of color channels
- the second color channel is One of a plurality of color channels, the distribution of the second color channel target pixel points describing the number of pixel points in the second color channel having pixel values within each set threshold range, the set threshold The range is M, and M is a positive integer greater than or equal to 1;
- the identifying module 58 is configured to determine the target according to a preset algorithm according to the determined first color feature corresponding to the plurality of first different regions and the second color feature corresponding to the plurality of second different regions The body color of the vehicle.
- the processor 610 includes a memory 620 connected to the processor, and the memory 620 and the processor 610 are connected to each other through a bus 600. among them:
- a memory 620 configured to store program code executed by the processor
- the processor 610 is configured to execute the program code stored by the memory, and perform the following process:
- the first color feature corresponding to the plurality of different regions, the first color feature comprising at least one of a distribution of a first color channel target pixel point and a color moment of the first color channel, the first
- the second image includes a plurality of color channels, the first color channel is one of the plurality of color channels, and the distribution of the first color channel target pixel points describes the pixel values in the first color channel in each Setting a number of pixels in the threshold range, the set threshold range is N, and N is a positive integer greater than or equal to 1;
- the first image is subjected to a histogram equalization operation on the V channel of the HSV color space to obtain a second image, where the first image includes image information of the target vehicle; and the setting window is swept in the set order.
- the processor 610 further performs the following process: determining the first color channel target pixel corresponding to the plurality of different regions.
- the distribution of points is input to a linear support vector machine that determines the body color of the target vehicle based on a first calculation model corresponding to the distribution of the target pixel points of the first color channel.
- the processor 610 specifically performs the following process: determining the color of the first color channel corresponding to the plurality of different regions determined The moment is input to a linear support vector machine that determines the body color of the target vehicle based on a second calculation model corresponding to the color moment of the first color channel.
- the processor 610 specifically performs the following processes: inputting the distribution of the first color channel target pixel point corresponding to the determined plurality of different regions and the color moment of the first color channel, and inputting To a linear support vector machine, the linear support vector machine determines the target vehicle according to a third calculation model corresponding to a distribution of the first color channel target pixel point and a color moment of the first color channel the color of car.
- embodiments of the present invention can be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment, or a combination of software and hardware. Moreover, the invention can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) including computer usable program code.
- computer-usable storage media including but not limited to disk storage, CD-ROM, optical storage, etc.
- the computer program instructions can also be stored in a computer readable memory that can direct a computer or other programmable data processing device to operate in a particular manner, such that the instructions stored in the computer readable memory produce an article of manufacture comprising the instruction device.
- the apparatus implements the functions specified in one or more blocks of a flow or a flow and/or block diagram of the flowchart.
- These computer program instructions can also be loaded onto a computer or other programmable data processing device such that a series of operational steps are performed on a computer or other programmable device to produce computer-implemented processing for execution on a computer or other programmable device.
- the instructions provide steps for implementing the functions specified in one or more of the flow or in a block or blocks of a flow diagram.
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Data Mining & Analysis (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Engineering & Computer Science (AREA)
- Bioinformatics & Computational Biology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Analysis (AREA)
- Image Processing (AREA)
Abstract
一种车身颜色识别的方法及装置,用于解决车身受光照的影响很大时,车身颜色识别的准确率低的问题。该方法包括:将第一图像在色调饱和度亮度HSV颜色空间的亮度V通道上进行直方图均衡化操作得到第二图像,其中,所述第一图像中包含目标车辆的图像信息(S11),按照设定顺序滑动设定窗口,将所述第二图像划分为多个不同区域(S12),确定出所述多个不同区域对应的第一颜色特征(S13),根据确定出的多个不同区域对应的所述第一颜色特征,按照预设算法,确定出所述目标车辆的车身颜色(S14)。由于第一图像在V通道上进行了直方图均衡化操作,并根据确定出的不同第一颜色特征对车身颜色进行识别,提高了车身颜色识别的准确度。
Description
本申请要求于2016年09月05日提交中国专利局、申请号为201610803092.6、申请名称为“一种车身颜色识别的方法及装置”的中国专利申请的优先权,其全部内容通过引用结合在本申请中。
本发明涉及计算机视觉领域,尤其涉及一种车身颜色识别的方法及装置。
随着车辆数量和交通出行量的不断增长,在平安城市建设的智能监控领域中车身颜色识别尤为重要,当车辆的车牌信息难以识别、车款信息不易辨认时,车身颜色信息作为最为明显的特征,对于案件侦查、刑事追踪、套牌识别等方面有着举足轻重的作用。
由于车身颜色是金属色,在智能监控场景中自动识别车身颜色时,受光照的影响很大,识别出的车身颜色准确率低。
发明内容
本发明的目的是提供一种车身颜色识别的方法及装置,以解决车身受光照的影响很大时,车身颜色识别的准确率低的问题。
第一方面,本发明实施例提出一种车身颜色识别的方法,该方法包括:获取到包含目标车辆的第一图像,将第一图像在色调饱和度亮度(Hue Saturation Value,HSV)颜色空间的亮度V通道上进行直方图均衡化操作,确定出所述直方图均衡化操作后的第二图像;通过按照设定顺序滑动设定窗口的方式,将所述第二图像划分为多个不同区域,确定出所述第二图像中包含的多个不同区域分别对应的第一颜色特征;根据确定出的不同所述第一颜色特征,根据确定出的多个不同区域对应的所述第一颜色特征,按照预设算法,确定出所述目标车辆的车身颜色。
本发明实施例中,所述第二图像包含多种颜色空间,所述第一颜色特征包括第一颜色通道目标像素点的分布情况和所述第一颜色通道的颜色矩中的至少一个,所述第二图像中包含多个颜色通道,所述第一颜色通道为所述多个颜色通道中的一个,所述第一颜色通道目标像素点的分布情况描述了第一颜色通道中像素值在每一个设定阈值范围内的像素点的个数,所述设定阈值范围为N个,N为大于或等于1的正整数,本发明实施例中,N优先4以上的正整数。
本发明实施例将第一图像在V通道上进行了直方图均衡化操作,去除了光照对车身颜色的影响,并根据确定出的不同第一颜色特征对车身颜色进行识别,提高了车身颜色识别的准确度。
在一种可能的设计中,若所述第一颜色特征为第一颜色通道目标像素点的分布情况时,将确定出的多个不同区域对应的所述第一颜色通道目标像素点的分布情况,输入到线性支持向量机,所述线性支持向量机根据与所述第一颜色通道目标像素点的分布情况对应的第一计算模型,确定出所述目标车辆的车身颜色。
本发明实施例根据确定出的所述第一颜色通道目标像素点的分布情况对车身颜色进行识别,提高了车身颜色识别的准确度。
在一种可能的设计中,若所述第一颜色特征为所述第一颜色通道的颜色矩时,将确定出的多个不同区域对应的所述第一颜色通道的颜色矩,输入到线性支持向量机,所述线性支持向量机根据与所述第一颜色通道的颜色矩对应的第二计算模型,确定出所述目标车辆的车身颜色,其中,所述颜色矩包括一阶矩、二阶矩和三阶矩。
本发明实施例根据确定出的所述第一颜色通道的颜色矩对车身颜色进行识别,提高了车身颜色识别的准确度。
在一种可能的设计中,若所述第一颜色特征为第一颜色通道目标像素点的分布情况和所述第一颜色通道的颜色矩时,将确定出的多个不同区域对应的所述第一颜色通道目标像素点的分布情况和所述第一颜色通道的颜色矩,输入到线性支持向量机,所述线性支持向量机根据与所述第一颜色通道目标像素点的分布情况和所述第一颜色通道的颜色矩对应的第三计算模型,确定出所述目标车辆的车身颜色。
本发明实施例根据确定出的所述第一颜色通道目标像素点的分布情况和所述第一颜色通道的颜色矩对车身颜色进行识别,提高了车身颜色识别的准确度。
在一种可能的设计中,将第一图像在HSV颜色空间的V通道上进行直方图均衡化操作之前,还包括:
将所述第一图像调整至设定尺寸。
确定出设定尺寸的第一图像,提高了线性支持向量机处理数据的准确性。
在一种可能的设计中,所述第二图像的多个区域中相邻两个区域之间部分重叠。
第二图像中的多个区域中相邻两个区域之间部分重叠,确定第一颜色特征时,确保了第二图像中的每个像素点都进行了统计,提高了车身颜色识别的准确度。
第二方面,本发明实施例提出一种车身颜色识别的方法,该方法包括:获取到包含目标车辆的第一图像,将第一图像在色调饱和度亮度HSV颜色空间的亮度V通道上进行直方图均衡化操作,确定出所述直方图均衡化操作后的第二图像;通过按照设定顺序滑动设定窗口的方式,将所述第二图像划分为多个第一不同区域,确定出所述第二图像中包含的多个第一不同区域分别对应的第一颜色特征;根据预设规则在所述第一图像中选取所述目标车辆的局部图像信息的至少一个区域,将所述至少一个区域在HSV颜色空间的V通道上进行直方图均衡化操作,确定出所述直方图均衡化操作后的至少一个第三图像;按照设定顺序滑动设定窗口,将所述第三图像划分为多个第二不同区域,确定出所述多个第二不同区域对应的第二颜色特征,根据确定出的多个第一不同区域对应的所述第一颜色特征与多个第二不同区域对应的所述第二颜色特征,按照预设算法,确定出所述目标车辆的车身颜色。
本发明实施例中,所述第二图像包含多种颜色空间,所述第一颜色特征包括第一颜色通道目标像素点的分布情况和所述第一颜色通道的颜色矩中的至少一个,所述第二图像中包含多个颜色通道,所述第一颜色通道为所述多个颜色通道中的一个,所述第一颜色通道目标像素点的分布情况描述了第一颜色通道中像素值在每一个设定阈值范围内的像素点的个数,所述设定阈值范围为N个,N为大于或等于1的正整数;所述第二颜色特征包括第二颜色通道目标像素点的分布情况和所述第二颜色通道的颜色矩中的至少一个,所述第三图像中包含多个颜色通道,所述第二颜色通道为所述多个
颜色通道中的一个,所述第二颜色通道目标像素点的分布情况描述了第二颜色通道中像素值在每一个设定阈值范围内的像素点的个数,所述设定阈值范围为M个,M为大于或等于1的正整数;本发明实施例中,N和M优先4以上的正整数。
本发明实施例将第一图像以及第三图像在V通道上进行了直方图均衡化操作,去除了光照对车身颜色的影响,并根据确定出的不同第一颜色特征和不同第二颜色特征对车身颜色进行识别,提高了车身颜色识别的准确度。
在一种可能的设计中,若所述第一颜色特征为第一颜色通道目标像素点的分布情况,所述第二颜色特征为第二颜色通道目标像素点的分布情况时,将确定出的多个第一不同区域对应的所述第一颜色通道目标像素点的分布情况和多个第二不同区域对应的所述第二颜色通道目标像素点的分布情况,输入到线性支持向量机,所述线性支持向量机根据与所述第一颜色通道目标像素点的分布情况和所述第二颜色通道目标像素点的分布情况对应的第一计算模型,确定出所述目标车辆的车身颜色。
本发明实施例根据确定出的所述第一颜色通道目标像素点的分布情况和所述第二颜色通道目标像素点的分布情况对车身颜色进行识别,提高了车身颜色识别的准确度。
在一种可能的设计中,若所述第一颜色特征为所述第一颜色通道的颜色矩,所述第二颜色特征为所述第二颜色通道的颜色矩时,将确定出的多个第一不同区域对应的所述第一颜色通道的颜色矩和多个第二不同区域对应的所述第一颜色通道的颜色矩,输入到线性支持向量机,所述线性支持向量机根据与所述第一颜色通道的颜色矩和所述第二颜色通道的颜色矩对应的第二计算模型,确定出所述目标车辆的车身颜色,其中,所述颜色矩包括一阶矩、二阶矩和三阶矩。
本发明实施例根据确定出的所述第一颜色通道的颜色矩和所述第二颜色通道的颜色矩对车身颜色进行识别,提高了车身颜色识别的准确度。
在一种可能的设计中,若所述第一颜色特征为第一颜色通道目标像素点的分布情况和所述第一颜色通道的颜色矩、所述第二颜色特征为第二颜色通道目标像素点的分布情况和所述第二颜色通道的颜色矩时,将确定出的多个第一不同区域对应的所述第一颜色通道目标像素点的分布情况和所述第一颜色通道的颜色矩、多个第二不同区域对应的所述第二颜色通道目标像素点的分布情况和所述第二颜色通道的颜色矩,输入到线性支持向量机,所述线性支持向量机根据与所述第一颜色通道目标像素点的分布情况和所述第一颜色通道的颜色矩、以及所述第二颜色通道目标像素点的分布情况和所述第二颜色通道的颜色矩对应的第三计算模型,确定出所述目标车辆的车身颜色。
本发明实施例根据确定出的所述第一颜色通道目标像素点的分布情况和所述第一颜色通道的颜色矩对车身颜色进行识别、以及所述第二颜色通道目标像素点的分布情况和所述第二颜色通道的颜色矩,提高了车身颜色识别的准确度。
在一种可能的设计中,所述第三图像的多个第二不同区域中相邻两个区域之间部分重叠。
第三图像中的多个第二不同区域中相邻两个区域之间部分重叠,确定第一颜色特征时,确保了第三图像中的每个像素点都进行了统计,提高了车身颜色识别的准确度。
第三方面,本发明实施例提出一种车身颜色识别的装置,该装置包括:
处理模块,用于将第一图像在色调饱和度亮度HSV颜色空间的亮度V通道上进行直方图均衡化操作得到第二图像,其中,所述第一图像中包含目标车辆的图像信息;
划分模块,用于按照设定顺序滑动设定窗口,将所述第二图像划分为多个不同区域;
确定模块,用于确定出所述多个不同区域对应的第一颜色特征,所述第一颜色特征包括第一颜色通道目标像素点的分布情况和所述第一颜色通道的颜色矩中的至少一个,所述第二图像中包含多个颜色通道,所述第一颜色通道为所述多个颜色通道中的一个,所述第一颜色通道目标像素点的分布情况描述了第一颜色通道中像素值在每一个设定阈值范围内的像素点的个数,所述设定阈值范围为N个,N为大于或等于1的正整数;
识别模块,用于根据确定出的多个不同区域对应的所述第一颜色特征,按照预设算法,确定出所述目标车辆的车身颜色。
在一种可能的设计中,若所述第一颜色特征为第一颜色通道目标像素点的分布情况时,所述识别模块用于:
将确定出的多个不同区域对应的所述第一颜色通道目标像素点的分布情况,输入到线性支持向量机,所述线性支持向量机根据与所述第一颜色通道目标像素点的分布情况对应的第一计算模型,确定出所述目标车辆的车身颜色。
在一种可能的设计中,若所述第一颜色特征为所述第一颜色通道的颜色矩时,所述识别模块用于:
将确定出的多个不同区域对应的所述第一颜色通道的颜色矩,输入到线性支持向量机,所述线性支持向量机根据与所述第一颜色通道的颜色矩对应的第二计算模型,确定出所述目标车辆的车身颜色。
在一种可能的设计中,若所述第一颜色特征为第一颜色通道目标像素点的分布情况和所述第一颜色通道的颜色矩时,所述识别模块用于:
将确定出的多个不同区域对应的所述第一颜色通道目标像素点的分布情况和所述第一颜色通道的颜色矩,输入到线性支持向量机,所述线性支持向量机根据与所述第一颜色通道目标像素点的分布情况和所述第一颜色通道的颜色矩对应的第三计算模型,确定出所述目标车辆的车身颜色。
第四方面,本发明实施例提出一种车身颜色识别的装置,该装置包括:
第一处理模块,用于将第一图像在色调饱和度亮度HSV颜色空间的亮度V通道上进行直方图均衡化操作得到第二图像,其中,所述第一图像中包含目标车辆的图像信息;
第一划分模块,用于按照设定顺序滑动设定窗口,将所述第二图像划分为多个第一不同区域;
第一确定模块,用于确定出所述多个第一不同区域对应的第一颜色特征,所述第一颜色特征包括第一颜色通道目标像素点的分布情况和所述第一颜色通道的颜色矩中的至少一个,所述第二图像中包含多个颜色通道,所述第一颜色通道为所述多个颜色通道中的一个,所述第一颜色通道目标像素点的分布情况描述了第一颜色通道中像素值在每一个设定阈值范围内的像素点的个数,所述设定阈值范围为N个,N为大于或等于1的正整数;
获取模块,用于根据预设规则在所述第一图像中选取至少一个区域,所述至少一个区域中包含所述目标车辆的局部图像信息;
第二处理模块,用于将所述至少一个区域在HSV颜色空间的V通道上进行直方图均衡化操作得到至少一个第三图像;
第二划分模块,用于按照设定顺序滑动设定窗口,将所述第三图像划分为多个第二不同区域;
第二确定模块,用于确定出所述多个第二不同区域对应的第二颜色特征,所述第二颜色特征包括第二颜色通道目标像素点的分布情况和所述第二颜色通道的颜色矩中的至少一个,所述第三图像中包含多个颜色通道,所述第二颜色通道为所述多个颜色通道中的一个,所述第二颜色通道目标像素点的分布情况描述了第二颜色通道中像素值在每一个设定阈值范围内的像素点的个数,所述设定阈值范围为M个,M为大于或等于1的正整数;
识别模块,用于根据确定出的多个第一不同区域对应的所述第一颜色特征与多个第二不同区域对应的所述第二颜色特征,按照预设算法,确定出所述目标车辆的车身颜色。
第五方面,本发明实施例提出一种车身颜色识别装置,包括处理器、以及与该处理器连接的存储器,其中:
存储器,所述存储器用于存储所述处理器所执行的程序代码;
处理器,用于执行所述存储器所存储的程序代码,执行下列过程:
将第一图像在色调饱和度亮度HSV颜色空间的亮度V通道上进行直方图均衡化操作得到第二图像,其中,所述第一图像中包含目标车辆的图像信息;
按照设定顺序滑动设定窗口,将所述第二图像划分为多个不同区域;
确定出所述多个不同区域对应的第一颜色特征,所述第一颜色特征包括第一颜色通道目标像素点的分布情况和所述第一颜色通道的颜色矩中的至少一个,所述第二图像中包含多个颜色通道,所述第一颜色通道为所述多个颜色通道中的一个,所述第一颜色通道目标像素点的分布情况描述了第一颜色通道中像素值在每一个设定阈值范围内的像素点的个数,所述设定阈值范围为N个,N为大于或等于1的正整数;
根据确定出的多个不同区域对应的所述第一颜色特征,按照预设算法,确定出所述目标车辆的车身颜色。
相较于现有技术,本发明提供的方案提高了车身颜色识别的准确度。
图1为本发明实施例提供的一种车身颜色识别方法的流程示意图;
图2为本发明实施例提供的又一种车身颜色识别方法的流程示意图;
图3为本发明实施例提供的另一种车身颜色识别方法的流程示意图;
图4为本发明实施例提供的一种车身颜色识别装置的结构示意图;
图5为本发明实施例提供的又一种车身颜色识别装置的结构示意图;
图6为本发明实施例提供的一种车身颜色识别装置的硬件结构示意图。
下面结合说明书附图对本发明实施例作进一步详细描述。应当理解,此处所描述
的实施例仅用于说明和解释本发明,并不用于限定本发明。
在智能监控场景中,当从监控设备中获取到图像中的目标车辆的车牌信息难以识别、车款信息不易辨认时,需要对目标车辆的颜色需要进行自动识别。
本发明实施例提供了一种车身颜色识别的方法,如图1所示,该方法包括以下过程:
S11、处理器将第一图像在色调饱和度亮度HSV颜色空间的亮度V通道上进行直方图均衡化操作得到第二图像,如图2所示,其中,所述HSV颜色空间是一种六角锥体模型,该模型是按照色调饱和度亮度来描述的,常应用于图像处理方面。
可选的,在步骤S11之前,还包括处理器将获取到的包含目标车辆信息的第一图像调整至设定尺寸。
具体的,确定出第一图像中红绿蓝(Red Green Blue,RGB)颜色空间中各通道的取值,将RGB颜色空间各通道的取值转化为HSV颜色空间中各通道的取值,具体如下:
v′=max
其中,r'为RGB颜色空间中R通道的值,g′为RGB颜色空间中G通道的值,b'为RGB颜色空间中B通道的值,h'为HSV颜色空间中H通道的值,s'为HSV颜色空间中S通道的值,v′为HSV颜色空间中V通道的值,max为RGB颜色空间中R,G,B三个通道中的最大值,min为RGB颜色空间中R,G,B三个通道中的最小值。
然后将h',s',v'归一化到[0,255],其中,
s=s′×255
v=v′
h',s',v'归一化为h,s,v之后,固定H通道和S通道,对V通道进行直方图均衡化操作,具体如下:
首先,统计出V通道中相同v值的像素点的个数的直方图,其中,v值为[0,255],
共256个级别;
然后,对每个v值的直方图(NumPixel),计算v值的分布密度(ProbPixel):
其中,i表示v值的取;H,W分别为第二图像的高和宽,H×W表示第二图像中像素点的总数。
再次,计算v值的累计直方图分布(CumuPixel):
最后,对v值进行均衡化操作:
v(i)=Cumupixel(i)×255
确定出HSV颜色空间的V通道均衡化操作后的v(i)后,将均衡化造作后的HSV颜色空间的转化为新的RGB颜色空间,具体如下:
p=v(i)×(1-s)
q=v(i)×(1-f×s)
t=v(i)×(1-(1-f)×s)
其中,r为新的RGB颜色空间中R通道的值,g为RGB颜色空间中G通道的值,b为新的RGB颜色空间中B通道的值。
本发明实施例中,所述RGB颜色空间是一种三维立方体模型,该模型是按照物理三基色描述的,常应用于图像处理方面。
S12、处理器按照设定顺序滑动设定窗口,将所述第二图像划分为多个不同区域;
本发明实施例中,还可以通过其它方式将第二图像划分为多个区域,本发明对其不做限定。
可选的,所述第二图像的多个区域中相邻两个区域之间部分重叠。
S13、确定出所述多个不同区域对应的第一颜色特征,如上图2所示。
本发明实施例中,所述第一颜色特征包括第一颜色通道目标像素点的分布情况和所述第一颜色通道的颜色矩中的至少一个,所述第二图像中包含多个颜色通道,所述第一颜色通道为所述多个颜色通道中的一个,所述第一颜色通道目标像素点的分布情况描述了第一颜色通道中像素值在每一个设定阈值范围内的像素点的个数,所述设定
阈值范围为N个,N为大于或等于1的正整数。
本发明实施例中,每个图像都可以采用多种不同的颜色空间表示,每个颜色空间对应多条颜色通道,即每个图像都有多个颜色通道,所述颜色通道为保存图像颜色信息的通道。
具体的,确定出所述第二图像划分出的每个区域的至少一种颜色空间中各通道的像素值在不同阈值范围内像素点的个数,作为第一颜色特征,其中,所述阈值范围至少为4个,即将0~255,分成4个连续的阈值范围,本发明实施例中,对阈值范围的划分不做限定,具体由实际需要确定。
举例说明,将0~255分成4个连续的阈值范围,所述4个连续的阈值范围分别为0~63、64~127、128~190、191~255。
还可以将第二图像划分出的每个区域的至少一种颜色空间中各通道的颜色矩,作为第一颜色特征。
也可以把确定出所述第二图像划分出的每个区域的至少一种颜色空间中各通道的像素值在不同阈值范围内像素点的个数,和第二图像划分出的每个区域的至少一种颜色空间中各通道的颜色矩,同时作为第一颜色特征。
本发明实施例中,在确定颜色特征时,还可以采用其它颜色空间,本发明实施例对其不做限定。
本发明实施例中,对任一颜色通道的颜色矩的分为一阶矩、二阶矩和三阶矩,具体计算如下:
其中pj,i表示第j个颜色通道中值为i的像素点出现的次数,N表示为对应区域中的像素点的总个数。
S14、根据确定出的多个不同区域对应的所述第一颜色特征,按照预设算法,确定出所述目标车辆的车身颜色,如上图2所示。
可选的,处理器将确定出的不同第一颜色特征处理器输入至预先训练好的线性支持向量机,确定出所述目标车辆的车身颜色。
本发明实施例中,输入的第一颜色特征为不同的数据时,线性支持向量机对应的计算模型不同。
本发明实施例中,将第一图像在HSV颜色空间的V通道上进行直方图均衡化操作得到第二图像,所述第一图像中包含目标车辆的图像信息;按照设定顺序滑动设定窗口,将所述第二图像划分为多个不同区域;确定出所述第二图像中包含的多个不同区域分别对应的第一颜色特征;根据确定出的不同所述第一颜色特征,确定出所述目标车辆的车身颜色。由于第一图像在V通道上进行了直方图均衡化操作,并根据确定出的不同第一颜色特征对车身颜色进行识别,提高了车身颜色识别的准确度。
本发明实施例中,如图3所示,处理器在调整至设定尺寸的第一图像中,采用类
金字塔的方式确定出包含所述目标车辆局部信息的一个区域,将该包含所述目标车辆局部信息的区域在HSV颜色空间的V通道上进行直方图均衡化操作得到第三图像,具体的均衡化操作过程与上述步骤S11一致,在此不对其进行赘述。
通过滑动窗口的方式,将所述第三图像划分为多个区域,确定出所述第三图像中包含的多个不同区域分别对应的第二颜色特征,其中,所述第三图像的多个区域中相邻两个区域之间部分重叠。
具体的,确定出所述第三图像划分出的每个区域的至少一种颜色空间中各通道的像素值在不同阈值范围内像素点的个数,作为第二颜色特征,其中,所述阈值范围至少为4个,即将0~255,分成4个阈值范围,本发明实施例中,对阈值范围的划分不做限定,具体由实际需要确定。
还可以将第三图像划分出的每个区域的至少一种颜色空间中各通道的颜色矩,作为第二颜色特征。
也可以把确定出所述第三图像划分出的每个区域的至少一种颜色空间中各通道的像素值在不同阈值范围内像素点的个数,和第三图像划分出的每个区域的至少一种颜色空间中各通道的颜色矩,同时作为第二颜色特征。
第二颜色特征中颜色矩的计算方法与上述第一颜色特征中颜色矩的计算方法一致,在此不对其进行赘述。
本发明实施例中,将确定出的不同第一颜色特征与不同第二颜色特征输入至预先训练好的线性支持向量机,确定出所述目标车辆的车身颜色。
同时使用第一颜色特征与第二颜色特征确定目标车辆的车身颜色,提高了车身颜色识别的准确度。
本发明实施例中,除了采用类金字塔的方式确定出包含所述目标车辆局部信息的一个区域,还可以采用其它的金字塔形式,例如将第二图像分成四个包含目标车辆信息的区域,然后对每个区域进行直方图均衡化处理后,分别提取该区域对应的第二颜色特征,本发明实施例中还可以采用其它方式对第二图像进行划分,本发明对其不做限定。
基于同一发明构思,本发明实施例提供的一种车身颜色识别的装置,如图4所示,该装置包括:
处理模块41,用于将第一图像在色调饱和度亮度HSV颜色空间的亮度V通道上进行直方图均衡化操作得到第二图像,其中,所述第一图像中包含目标车辆的图像信息;
划分模块42,用于按照设定顺序滑动设定窗口,将所述第二图像划分为多个不同区域;
确定模块43,用于确定出所述多个不同区域对应的第一颜色特征,所述第一颜色特征包括第一颜色通道目标像素点的分布情况和所述第一颜色通道的颜色矩中的至少一个,所述第二图像中包含多个颜色通道,所述第一颜色通道为所述多个颜色通道中的一个,所述第一颜色通道目标像素点的分布情况描述了第一颜色通道中像素值在每一个设定阈值范围内的像素点的个数,所述设定阈值范围为N个,N为大于或等于1的正整数;
识别模块44,用于根据确定出的多个不同区域对应的所述第一颜色特征,按照预
设算法,确定出所述目标车辆的车身颜色。
本发明实施例中,将第一图像在HSV颜色空间的V通道上进行直方图均衡化操作得到第二图像,所述第一图像中包含目标车辆的图像信息;按照设定顺序滑动设定窗口,将所述第二图像划分为多个不同区域;确定出所述第二图像中包含的多个不同区域分别对应的第一颜色特征;根据确定出的不同所述第一颜色特征,确定出所述目标车辆的车身颜色。由于第一图像在V通道上进行了直方图均衡化操作,并根据确定出的不同第一颜色特征对车身颜色进行识别,提高了车身颜色识别的准确度。
可选的,若所述第一颜色特征为第一颜色通道目标像素点的分布情况时,所述识别模块用于:将确定出的多个不同区域对应的所述第一颜色通道目标像素点的分布情况,输入到线性支持向量机,所述线性支持向量机根据与所述第一颜色通道目标像素点的分布情况对应的第一计算模型,确定出所述目标车辆的车身颜色。
可选的,若所述第一颜色特征为所述第一颜色通道的颜色矩时,所述识别模块用于:将确定出的多个不同区域对应的所述第一颜色通道的颜色矩,输入到线性支持向量机,所述线性支持向量机根据与所述第一颜色通道的颜色矩对应的第二计算模型,确定出所述目标车辆的车身颜色。
可选的,若所述第一颜色特征为第一颜色通道目标像素点的分布情况和所述第一颜色通道的颜色矩时,所述识别模块用于:将确定出的多个不同区域对应的所述第一颜色通道目标像素点的分布情况和所述第一颜色通道的颜色矩,输入到线性支持向量机,所述线性支持向量机根据与所述第一颜色通道目标像素点的分布情况和所述第一颜色通道的颜色矩对应的第三计算模型,确定出所述目标车辆的车身颜色。
基于同一发明构思,本发明实施例提供的一种车身颜色识别的装置,如图5所示,该装置包括:
第一处理模块51,用于将第一图像在色调饱和度亮度HSV颜色空间的亮度V通道上进行直方图均衡化操作得到第二图像,其中,所述第一图像中包含目标车辆的图像信息;
第一划分模块52,用于按照设定顺序滑动设定窗口,将所述第二图像划分为多个第一不同区域;
第一确定模块53,用于确定出所述多个第一不同区域对应的第一颜色特征,所述第一颜色特征包括第一颜色通道目标像素点的分布情况和所述第一颜色通道的颜色矩中的至少一个,所述第二图像中包含多个颜色通道,所述第一颜色通道为所述多个颜色通道中的一个,所述第一颜色通道目标像素点的分布情况描述了第一颜色通道中像素值在每一个设定阈值范围内的像素点的个数,所述设定阈值范围为N个,N为大于或等于1的正整数;
获取模块54,用于根据预设规则在所述第一图像中选取至少一个区域,所述至少一个区域中包含所述目标车辆的局部图像信息;
第二处理模块55,用于将所述至少一个区域在HSV颜色空间的V通道上进行直方图均衡化操作得到至少一个第三图像;
第二划分模块56,用于按照设定顺序滑动设定窗口,将所述第三图像划分为多个第二不同区域;
第二确定模块57,用于确定出所述多个第二不同区域对应的第二颜色特征,所述
第二颜色特征包括第二颜色通道目标像素点的分布情况和所述第二颜色通道的颜色矩中的至少一个,所述第三图像中包含多个颜色通道,所述第二颜色通道为所述多个颜色通道中的一个,所述第二颜色通道目标像素点的分布情况描述了第二颜色通道中像素值在每一个设定阈值范围内的像素点的个数,所述设定阈值范围为M个,M为大于或等于1的正整数;
识别模块58,用于根据确定出的多个第一不同区域对应的所述第一颜色特征与多个第二不同区域对应的所述第二颜色特征,按照预设算法,确定出所述目标车辆的车身颜色。
下面结合优选的硬件结构,对本发明实施例提供的装置的结构、处理方式进行说明。
本发明实施例提出一种车身颜色识别系统,如图6所示,包括处理器610、以及与该处理器连接的存储器620,所述存储器620和所述处理器610分别通过总线600相互连接,其中:
存储器620,用于存储所述处理器所执行的程序代码;
处理器610,用于用于执行所述存储器所存储的程序代码,执行下列过程:
将第一图像在色调饱和度亮度HSV颜色空间的亮度V通道上进行直方图均衡化操作得到第二图像,其中,所述第一图像中包含目标车辆的图像信息;
按照设定顺序滑动设定窗口,将所述第二图像划分为多个不同区域;
确定出所述多个不同区域对应的第一颜色特征,所述第一颜色特征包括第一颜色通道目标像素点的分布情况和所述第一颜色通道的颜色矩中的至少一个,所述第二图像中包含多个颜色通道,所述第一颜色通道为所述多个颜色通道中的一个,所述第一颜色通道目标像素点的分布情况描述了第一颜色通道中像素值在每一个设定阈值范围内的像素点的个数,所述设定阈值范围为N个,N为大于或等于1的正整数;
根据确定出的多个不同区域对应的所述第一颜色特征,按照预设算法,确定出所述目标车辆的车身颜色。
本发明实施例中,将第一图像在HSV颜色空间的V通道上进行直方图均衡化操作得到第二图像,所述第一图像中包含目标车辆的图像信息;按照设定顺序滑动设定窗口,将所述第二图像划分为多个不同区域;确定出所述第二图像中包含的多个不同区域分别对应的第一颜色特征;根据确定出的不同所述第一颜色特征,确定出所述目标车辆的车身颜色。由于第一图像在V通道上进行了直方图均衡化操作,并根据确定出的不同第一颜色特征对车身颜色进行识别,提高了车身颜色识别的准确度。
可选的,若所述第一颜色特征为第一颜色通道目标像素点的分布情况时,处理器610还执行下列过程:将确定出的多个不同区域对应的所述第一颜色通道目标像素点的分布情况,输入到线性支持向量机,所述线性支持向量机根据与所述第一颜色通道目标像素点的分布情况对应的第一计算模型,确定出所述目标车辆的车身颜色。
可选的,若所述第一颜色特征为所述第一颜色通道的颜色矩时,处理器610具体还执行下列过程:将确定出的多个不同区域对应的所述第一颜色通道的颜色矩,输入到线性支持向量机,所述线性支持向量机根据与所述第一颜色通道的颜色矩对应的第二计算模型,确定出所述目标车辆的车身颜色。
可选的,若所述第一颜色特征为第一颜色通道目标像素点的分布情况和所述第一
颜色通道的颜色矩时,处理器610具体还执行下列过程:将确定出的多个不同区域对应的所述第一颜色通道目标像素点的分布情况和所述第一颜色通道的颜色矩,输入到线性支持向量机,所述线性支持向量机根据与所述第一颜色通道目标像素点的分布情况和所述第一颜色通道的颜色矩对应的第三计算模型,确定出所述目标车辆的车身颜色。
本领域内的技术人员应明白,本发明的实施例可提供为方法、系统、或计算机程序产品。因此,本发明可采用完全硬件实施例、完全软件实施例、或结合软件和硬件方面的实施例的形式。而且,本发明可采用在一个或多个其中包含有计算机可用程序代码的计算机可用存储介质(包括但不限于磁盘存储器、CD-ROM、光学存储器等)上实施的计算机程序产品的形式。
本发明是参照根据本发明实施例的方法、设备(系统)、和计算机程序产品的流程图和/或方框图来描述的。应理解可由计算机程序指令实现流程图和/或方框图中的每一流程和/或方框、以及流程图和/或方框图中的流程和/或方框的结合。可提供这些计算机程序指令到通用计算机、专用计算机、嵌入式处理机或其他可编程数据处理设备的处理器以产生一个机器,使得通过计算机或其他可编程数据处理设备的处理器执行的指令产生用于实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的装置。
这些计算机程序指令也可存储在能引导计算机或其他可编程数据处理设备以特定方式工作的计算机可读存储器中,使得存储在该计算机可读存储器中的指令产生包括指令装置的制造品,该指令装置实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能。
这些计算机程序指令也可装载到计算机或其他可编程数据处理设备上,使得在计算机或其他可编程设备上执行一系列操作步骤以产生计算机实现的处理,从而在计算机或其他可编程设备上执行的指令提供用于实现在流程图一个流程或多个流程和/或方框图一个方框或多个方框中指定的功能的步骤。
尽管已描述了本发明的优选实施例,但本领域内的技术人员一旦得知了基本创造性概念,则可对这些实施例作出另外的变更和修改。所以,所附权利要求意欲解释为包括优选实施例以及落入本发明范围的所有变更和修改。
显然,本领域的技术人员可以对本发明进行各种改动和变型而不脱离本发明的精神和范围。这样,倘若本发明的这些修改和变型属于本发明权利要求及其等同技术的范围之内,则本发明也意图包含这些改动和变型在内。
Claims (11)
- 一种车身颜色识别的方法,其特征在于,该方法包括:将第一图像在色调饱和度亮度HSV颜色空间的亮度V通道上进行直方图均衡化操作得到第二图像,其中,所述第一图像中包含目标车辆的图像信息;按照设定顺序滑动设定窗口,将所述第二图像划分为多个不同区域;确定出所述多个不同区域对应的第一颜色特征,所述第一颜色特征包括第一颜色通道目标像素点的分布情况和所述第一颜色通道的颜色矩中的至少一个,所述第二图像中包含多个颜色通道,所述第一颜色通道为所述多个颜色通道中的一个,所述第一颜色通道目标像素点的分布情况描述了第一颜色通道中像素值在每一个设定阈值范围内的像素点的个数,所述设定阈值范围为N个,N为大于或等于1的正整数;根据确定出的多个不同区域对应的所述第一颜色特征,按照预设算法,确定出所述目标车辆的车身颜色。
- 如权利要求1所述的方法,其特征在于,若所述第一颜色特征为第一颜色通道目标像素点的分布情况时,根据确定出的多个不同区域对应的所述第一颜色特征,按照预设算法,确定出所述目标车辆的车身颜色,包括:将确定出的多个不同区域对应的所述第一颜色通道目标像素点的分布情况,输入到线性支持向量机,所述线性支持向量机根据与所述第一颜色通道目标像素点的分布情况对应的第一计算模型,确定出所述目标车辆的车身颜色。
- 如权利要求1所述的方法,其特征在于,若所述第一颜色特征为所述第一颜色通道的颜色矩时,根据确定出的多个不同区域对应的所述第一颜色特征,按照预设算法,确定出所述目标车辆的车身颜色,包括:将确定出的多个不同区域对应的所述第一颜色通道的颜色矩,输入到线性支持向量机,所述线性支持向量机根据与所述第一颜色通道的颜色矩对应的第二计算模型,确定出所述目标车辆的车身颜色。
- 如权利要求1所述的方法,其特征在于,若所述第一颜色特征为第一颜色通道目标像素点的分布情况和所述第一颜色通道的颜色矩时,根据确定出的多个不同区域对应的所述第一颜色特征,按照预设算法,确定出所述目标车辆的车身颜色,包括:将确定出的多个不同区域对应的所述第一颜色通道目标像素点的分布情况和所述第一颜色通道的颜色矩,输入到线性支持向量机,所述线性支持向量机根据与所述第一颜色通道目标像素点的分布情况和所述第一颜色通道的颜色矩对应的第三计算模型,确定出所述目标车辆的车身颜色。
- 一种车身颜色识别的方法,其特征在于,该方法包括:将第一图像在色调饱和度亮度HSV颜色空间的亮度V通道上进行直方图均衡化操作得到第二图像,其中,所述第一图像中包含目标车辆的图像信息;按照设定顺序滑动设定窗口,将所述第二图像划分为多个第一不同区域;确定出所述多个第一不同区域对应的第一颜色特征,所述第一颜色特征包括第一颜色通道目标像素点的分布情况和所述第一颜色通道的颜色矩中的至少一个,所述第二图像中包含多个颜色通道,所述第一颜色通道为所述多个颜色通道中的一个,所述第一颜色通道目标像素点的分布情况描述了第一颜色通道中像素值在每一个设定阈值 范围内的像素点的个数,所述设定阈值范围为N个,N为大于或等于1的正整数;根据预设规则在所述第一图像中选取至少一个区域,所述至少一个区域中包含所述目标车辆的局部图像信息;将所述至少一个区域在HSV颜色空间的V通道上进行直方图均衡化操作得到至少一个第三图像;按照设定顺序滑动设定窗口,将所述第三图像划分为多个第二不同区域;确定出所述多个第二不同区域对应的第二颜色特征,所述第二颜色特征包括第二颜色通道目标像素点的分布情况和所述第二颜色通道的颜色矩中的至少一个,所述第三图像中包含多个颜色通道,所述第二颜色通道为所述多个颜色通道中的一个,所述第二颜色通道目标像素点的分布情况描述了第二颜色通道中像素值在每一个设定阈值范围内的像素点的个数,所述设定阈值范围为M个,M为大于或等于1的正整数;根据确定出的多个第一不同区域对应的所述第一颜色特征与多个第二不同区域对应的所述第二颜色特征,按照预设算法,确定出所述目标车辆的车身颜色。
- 一种车身颜色识别的装置,其特征在于,该装置包括:处理模块,用于将第一图像在色调饱和度亮度HSV颜色空间的亮度V通道上进行直方图均衡化操作得到第二图像,其中,所述第一图像中包含目标车辆的图像信息;划分模块,用于按照设定顺序滑动设定窗口,将所述第二图像划分为多个不同区域;确定模块,用于确定出所述多个不同区域对应的第一颜色特征,所述第一颜色特征包括第一颜色通道目标像素点的分布情况和所述第一颜色通道的颜色矩中的至少一个,所述第二图像中包含多个颜色通道,所述第一颜色通道为所述多个颜色通道中的一个,所述第一颜色通道目标像素点的分布情况描述了第一颜色通道中像素值在每一个设定阈值范围内的像素点的个数,所述设定阈值范围为N个,N为大于或等于1的正整数;识别模块,用于根据确定出的多个不同区域对应的所述第一颜色特征,按照预设算法,确定出所述目标车辆的车身颜色。
- 如权利要求6所述的装置,其特征在于,若所述第一颜色特征为第一颜色通道目标像素点的分布情况时,所述识别模块用于:将确定出的多个不同区域对应的所述第一颜色通道目标像素点的分布情况,输入到线性支持向量机,所述线性支持向量机根据与所述第一颜色通道目标像素点的分布情况对应的第一计算模型,确定出所述目标车辆的车身颜色。
- 如权利要求6所述的装置,其特征在于,若所述第一颜色特征为所述第一颜色通道的颜色矩时,所述识别模块用于:将确定出的多个不同区域对应的所述第一颜色通道的颜色矩,输入到线性支持向量机,所述线性支持向量机根据与所述第一颜色通道的颜色矩对应的第二计算模型,确定出所述目标车辆的车身颜色。
- 如权利要求6所述的装置,其特征在于,若所述第一颜色特征为第一颜色通道目标像素点的分布情况和所述第一颜色通道的颜色矩时,所述识别模块用于:将确定出的多个不同区域对应的所述第一颜色通道目标像素点的分布情况和所述第一颜色通道的颜色矩,输入到线性支持向量机,所述线性支持向量机根据与所述第 一颜色通道目标像素点的分布情况和所述第一颜色通道的颜色矩对应的第三计算模型,确定出所述目标车辆的车身颜色。
- 一种车身颜色识别的装置,其特征在于,该装置包括:第一处理模块,用于将第一图像在色调饱和度亮度HSV颜色空间的亮度V通道上进行直方图均衡化操作得到第二图像,其中,所述第一图像中包含目标车辆的图像信息;第一划分模块,用于按照设定顺序滑动设定窗口,将所述第二图像划分为多个第一不同区域;第一确定模块,用于确定出所述多个第一不同区域对应的第一颜色特征,所述第一颜色特征包括第一颜色通道目标像素点的分布情况和所述第一颜色通道的颜色矩中的至少一个,所述第二图像中包含多个颜色通道,所述第一颜色通道为所述多个颜色通道中的一个,所述第一颜色通道目标像素点的分布情况描述了第一颜色通道中像素值在每一个设定阈值范围内的像素点的个数,所述设定阈值范围为N个,N为大于或等于1的正整数;获取模块,用于根据预设规则在所述第一图像中选取至少一个区域,所述至少一个区域中包含所述目标车辆的局部图像信息;第二处理模块,用于将所述至少一个区域在HSV颜色空间的V通道上进行直方图均衡化操作得到至少一个第三图像;第二划分模块,用于按照设定顺序滑动设定窗口,将所述第三图像划分为多个第二不同区域;第二确定模块,用于确定出所述多个第二不同区域对应的第二颜色特征,所述第二颜色特征包括第二颜色通道目标像素点的分布情况和所述第二颜色通道的颜色矩中的至少一个,所述第三图像中包含多个颜色通道,所述第二颜色通道为所述多个颜色通道中的一个,所述第二颜色通道目标像素点的分布情况描述了第二颜色通道中像素值在每一个设定阈值范围内的像素点的个数,所述设定阈值范围为M个,M为大于或等于1的正整数;识别模块,用于根据确定出的多个第一不同区域对应的所述第一颜色特征与多个第二不同区域对应的所述第二颜色特征,按照预设算法,确定出所述目标车辆的车身颜色。
- 一种车身颜色识别的装置,其特征在于,包括:处理器和存储器;所述存储器和所述处理器分别通过总线相互连接;所述存储器用于存储所述处理器所执行的程序代码;所述处理器用于执行所述存储器所存储的程序代码,具体用于执行权利要求1至5任一项所述的方法。
Applications Claiming Priority (2)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610803092.6A CN106485199A (zh) | 2016-09-05 | 2016-09-05 | 一种车身颜色识别的方法及装置 |
CN201610803092.6 | 2016-09-05 |
Publications (1)
Publication Number | Publication Date |
---|---|
WO2018040756A1 true WO2018040756A1 (zh) | 2018-03-08 |
Family
ID=58273459
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
PCT/CN2017/092651 WO2018040756A1 (zh) | 2016-09-05 | 2017-07-12 | 一种车身颜色识别的方法及装置 |
Country Status (2)
Country | Link |
---|---|
CN (1) | CN106485199A (zh) |
WO (1) | WO2018040756A1 (zh) |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110751053A (zh) * | 2019-09-26 | 2020-02-04 | 高新兴科技集团股份有限公司 | 一种车辆颜色识别方法、装置、设备及存储介质 |
CN112597840A (zh) * | 2020-12-14 | 2021-04-02 | 深圳集智数字科技有限公司 | 一种图像识别方法、装置及设备 |
CN113763491A (zh) * | 2021-08-26 | 2021-12-07 | 浙江中烟工业有限责任公司 | 一种烟丝桶残余物的视觉检测方法 |
CN115100196A (zh) * | 2022-08-24 | 2022-09-23 | 聊城市洛溪信息科技有限公司 | 一种基于图像分割的冲压件除锈效果评估方法 |
CN116416325A (zh) * | 2023-06-12 | 2023-07-11 | 广东粤港澳大湾区国家纳米科技创新研究院 | 一种颜色色块自动识别方法、装置及电子设备 |
CN116563770A (zh) * | 2023-07-10 | 2023-08-08 | 四川弘和数智集团有限公司 | 车辆颜色的检测方法、装置及设备和介质 |
CN116740383A (zh) * | 2023-06-19 | 2023-09-12 | 清远市保鸿涂料有限公司 | 一种基于图像识别的车漆配方精准识别方法及系统 |
Families Citing this family (9)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106485199A (zh) * | 2016-09-05 | 2017-03-08 | 华为技术有限公司 | 一种车身颜色识别的方法及装置 |
CN108563976B (zh) * | 2017-11-29 | 2021-04-02 | 浙江工业大学 | 一种基于车窗位置的多方向车辆颜色识别方法 |
CN108154502B (zh) * | 2017-12-22 | 2022-01-11 | 王华锋 | 一种基于卷积神经网络的通孔焊点识别方法 |
CN109146826A (zh) * | 2018-06-27 | 2019-01-04 | 华南理工大学 | 一种图像增强方法及装置 |
CN111626967A (zh) * | 2019-02-28 | 2020-09-04 | 富华科精密工业(深圳)有限公司 | 图像增强方法、装置、计算机装置及可读存储介质 |
CN109886896B (zh) * | 2019-02-28 | 2021-04-27 | 闽江学院 | 一种蓝色车牌分割与矫正方法 |
CN111833256A (zh) * | 2019-04-17 | 2020-10-27 | 富华科精密工业(深圳)有限公司 | 图像增强方法、装置、计算机装置及可读存储介质 |
CN111860533B (zh) * | 2019-04-30 | 2023-12-12 | 深圳数字生命研究院 | 图像的识别方法及装置、存储介质和电子装置 |
CN113570583B (zh) * | 2021-07-30 | 2024-06-07 | 北京市商汤科技开发有限公司 | 图像处理方法及装置、电子设备和存储介质 |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100098331A1 (en) * | 2008-09-26 | 2010-04-22 | Sony Corporation | System and method for segmenting foreground and background in a video |
CN102393902A (zh) * | 2011-12-01 | 2012-03-28 | 昆山市工业技术研究院有限责任公司 | 基于h_s二维直方图和区域颜色匹配的车辆颜色检测方法 |
CN103106386A (zh) * | 2011-11-10 | 2013-05-15 | 华为技术有限公司 | 动态自适应肤色分割方法和装置 |
CN104715239A (zh) * | 2015-03-12 | 2015-06-17 | 哈尔滨工程大学 | 一种基于去雾处理和权重分块的车辆颜色识别方法 |
CN105354530A (zh) * | 2015-09-22 | 2016-02-24 | 浙江宇视科技有限公司 | 一种车身颜色识别方法及装置 |
CN106485199A (zh) * | 2016-09-05 | 2017-03-08 | 华为技术有限公司 | 一种车身颜色识别的方法及装置 |
Family Cites Families (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2012147256A1 (ja) * | 2011-04-25 | 2012-11-01 | パナソニック株式会社 | 画像処理装置 |
CN103440348B (zh) * | 2013-09-16 | 2016-11-02 | 重庆邮电大学 | 一种基于矢量量化的全局和局部颜色图像检索方法 |
CN103544480A (zh) * | 2013-10-17 | 2014-01-29 | 公安部第三研究所 | 车辆颜色识别方法 |
CN105160691A (zh) * | 2015-08-29 | 2015-12-16 | 电子科技大学 | 基于颜色直方图的车身颜色识别方法 |
-
2016
- 2016-09-05 CN CN201610803092.6A patent/CN106485199A/zh active Pending
-
2017
- 2017-07-12 WO PCT/CN2017/092651 patent/WO2018040756A1/zh active Application Filing
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20100098331A1 (en) * | 2008-09-26 | 2010-04-22 | Sony Corporation | System and method for segmenting foreground and background in a video |
CN103106386A (zh) * | 2011-11-10 | 2013-05-15 | 华为技术有限公司 | 动态自适应肤色分割方法和装置 |
CN102393902A (zh) * | 2011-12-01 | 2012-03-28 | 昆山市工业技术研究院有限责任公司 | 基于h_s二维直方图和区域颜色匹配的车辆颜色检测方法 |
CN104715239A (zh) * | 2015-03-12 | 2015-06-17 | 哈尔滨工程大学 | 一种基于去雾处理和权重分块的车辆颜色识别方法 |
CN105354530A (zh) * | 2015-09-22 | 2016-02-24 | 浙江宇视科技有限公司 | 一种车身颜色识别方法及装置 |
CN106485199A (zh) * | 2016-09-05 | 2017-03-08 | 华为技术有限公司 | 一种车身颜色识别的方法及装置 |
Cited By (11)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110751053A (zh) * | 2019-09-26 | 2020-02-04 | 高新兴科技集团股份有限公司 | 一种车辆颜色识别方法、装置、设备及存储介质 |
CN110751053B (zh) * | 2019-09-26 | 2022-02-22 | 高新兴科技集团股份有限公司 | 一种车辆颜色识别方法、装置、设备及存储介质 |
CN112597840A (zh) * | 2020-12-14 | 2021-04-02 | 深圳集智数字科技有限公司 | 一种图像识别方法、装置及设备 |
CN113763491A (zh) * | 2021-08-26 | 2021-12-07 | 浙江中烟工业有限责任公司 | 一种烟丝桶残余物的视觉检测方法 |
CN113763491B (zh) * | 2021-08-26 | 2024-03-12 | 浙江中烟工业有限责任公司 | 一种烟丝桶残余物的视觉检测方法 |
CN115100196A (zh) * | 2022-08-24 | 2022-09-23 | 聊城市洛溪信息科技有限公司 | 一种基于图像分割的冲压件除锈效果评估方法 |
CN116416325A (zh) * | 2023-06-12 | 2023-07-11 | 广东粤港澳大湾区国家纳米科技创新研究院 | 一种颜色色块自动识别方法、装置及电子设备 |
CN116416325B (zh) * | 2023-06-12 | 2023-09-29 | 广东粤港澳大湾区国家纳米科技创新研究院 | 一种颜色色块自动识别方法、装置及电子设备 |
CN116740383A (zh) * | 2023-06-19 | 2023-09-12 | 清远市保鸿涂料有限公司 | 一种基于图像识别的车漆配方精准识别方法及系统 |
CN116563770A (zh) * | 2023-07-10 | 2023-08-08 | 四川弘和数智集团有限公司 | 车辆颜色的检测方法、装置及设备和介质 |
CN116563770B (zh) * | 2023-07-10 | 2023-09-29 | 四川弘和数智集团有限公司 | 车辆颜色的检测方法、装置及设备和介质 |
Also Published As
Publication number | Publication date |
---|---|
CN106485199A (zh) | 2017-03-08 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
WO2018040756A1 (zh) | 一种车身颜色识别的方法及装置 | |
US10275677B2 (en) | Image processing apparatus, image processing method and program | |
CN109918969B (zh) | 人脸检测方法及装置、计算机装置和计算机可读存储介质 | |
CN109918971B (zh) | 监控视频中人数检测方法及装置 | |
US8798361B2 (en) | Mapping colors of an image | |
US8295593B2 (en) | Method of detecting red-eye objects in digital images using color, structural, and geometric characteristics | |
US8103058B2 (en) | Detecting and tracking objects in digital images | |
KR20180065889A (ko) | 타겟의 검측 방법 및 장치 | |
US20210334998A1 (en) | Image processing method, apparatus, device and medium for locating center of target object region | |
CN110390643B (zh) | 一种车牌增强方法、装置及电子设备 | |
CN106446862A (zh) | 一种人脸检测方法及系统 | |
CN112101260B (zh) | 一种作业人员安全带识别方法、装置、设备和存储介质 | |
CN111145105B (zh) | 一种图像快速去雾方法、装置、终端及存储介质 | |
JP2020205118A (ja) | オブジェクト検出のためのシステム及び方法 | |
KR20120111153A (ko) | 번호판영역 검출을 위한 전처리 방법 및 장치 | |
US9672447B2 (en) | Segmentation based image transform | |
Katramados et al. | Real-time visual saliency by division of gaussians | |
CN111709305A (zh) | 一种基于局部图像块的人脸年龄识别方法 | |
EP3543954B1 (en) | Method of processing a video | |
Kuzovkin et al. | Descriptor-based image colorization and regularization | |
KR102010086B1 (ko) | 미세조직의 상 분할 방법 및 장치 | |
KR102118783B1 (ko) | 조명환경에 강인한 색상 필터링을 위한 전처리 방법 및 그 시스템 | |
TWI498830B (zh) | 不均勻光線下的車牌辨識方法及系統 | |
KR102149974B1 (ko) | 단일 영상의 안개 제거 연산량을 줄인 대기강도추정장치 및 그 장치의 구동방법 | |
CN104219423B (zh) | 一种信息处理方法和装置 |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
121 | Ep: the epo has been informed by wipo that ep was designated in this application |
Ref document number: 17845059 Country of ref document: EP Kind code of ref document: A1 |
|
NENP | Non-entry into the national phase |
Ref country code: DE |
|
122 | Ep: pct application non-entry in european phase |
Ref document number: 17845059 Country of ref document: EP Kind code of ref document: A1 |