WO2017054442A1 - 一种图像信息识别处理方法及装置、计算机存储介质 - Google Patents

一种图像信息识别处理方法及装置、计算机存储介质 Download PDF

Info

Publication number
WO2017054442A1
WO2017054442A1 PCT/CN2016/080564 CN2016080564W WO2017054442A1 WO 2017054442 A1 WO2017054442 A1 WO 2017054442A1 CN 2016080564 W CN2016080564 W CN 2016080564W WO 2017054442 A1 WO2017054442 A1 WO 2017054442A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
attribute value
sub
nth sub
detected
Prior art date
Application number
PCT/CN2016/080564
Other languages
English (en)
French (fr)
Inventor
郭德安
宋翔宇
Original Assignee
腾讯科技（深圳）有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 腾讯科技（深圳）有限公司 filed Critical 腾讯科技（深圳）有限公司
Priority to EP16850067.6A priority Critical patent/EP3273388A4/en
Publication of WO2017054442A1 publication Critical patent/WO2017054442A1/zh
Priority to US15/729,935 priority patent/US10438086B2/en

Links

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
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
    • G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/22—Matching criteria, e.g. proximity measures
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/20—Image preprocessing
    • G06V10/25—Determination of region of interest [ROI] or a volume of interest [VOI]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10016—Video; Image sequence
    • G06T2207/10021—Stereoscopic video; Stereoscopic image sequence
    • 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

Definitions

  • the present invention relates to image information processing technology, and in particular to an image information recognition processing method and apparatus, and a computer storage medium.
  • image similarity recognition methods include: color distribution histogram CF algorithm and perceptual hash phash algorithm.
  • the CF algorithm has a large error recognition probability, that is, it is easy to think that the image error that is not similar to the reference image is similar; the phash algorithm can guarantee the recognition accuracy rate to some extent, but it is easy to miss some similarities. Higher picture. It can be seen that there is a need for a solution that can ensure that the image with high similarity is not missed and that the recognition rate is guaranteed or improved.
  • embodiments of the present invention provide an image information recognition processing method and apparatus, and a computer storage medium, to at least solve the problems existing in the prior art, improve the recognition accuracy rate, and reduce the similar pictures being missed. Probability.
  • An embodiment of the present invention provides an image information recognition processing method, where the method includes:
  • first attribute values and/or values in the first reference image and the Nth sub-image respectively a second attribute value
  • the first attribute value is used to represent color distribution information of the image
  • the second attribute value is used to represent fingerprint information of the image
  • An embodiment of the present invention provides an image information recognition processing device, where the device includes:
  • a first acquiring unit configured to acquire a first reference image
  • a second acquiring unit configured to acquire at least one image to be detected, performing an Nth interception or cropping process on the image to be detected in at least one specified direction according to the first preset rule, to obtain the at least one image to be detected
  • the Nth sub-image, N is a positive integer
  • a third acquiring unit configured to respectively acquire a first attribute value and/or a second attribute value in the first reference image and the Nth sub-image, where the first attribute value is used to represent color distribution information of the image
  • the second attribute value is used to represent fingerprint information of the image
  • a first identifying unit configured to identify the Nth sub-image when a first attribute value and/or a second attribute value of the Nth sub-image and a corresponding attribute value of the first reference image satisfy a predetermined condition The to-be-detected image to which the member belongs is matched with the first reference image.
  • the embodiment of the present invention further provides a computer storage medium, wherein computer executable instructions are stored, and the computer executable instructions are configured to execute the image information recognition processing method.
  • the image information recognition processing method acquires a first reference image, acquires at least one image to be detected, and performs at least one N-th interception or cropping process on the image to be detected according to the first preset rule.
  • the similarity between the first attribute value and the second attribute value is obtained from the Nth sub-image and the first reference image, that is, both the color distribution of the image and the image fingerprint information are taken into consideration.
  • the effect of image similarity recognition; using the embodiment of the invention can effectively improve the recognition accuracy rate and reduce the probability that similar pictures are missed.
  • FIG. 1 is a schematic diagram of a hardware entity in which an image information recognition processing apparatus according to an embodiment of the present invention can be located;
  • FIG. 2 is a schematic flowchart of an implementation process according to Embodiment 1 of the present invention.
  • Embodiment 2 of the present invention is a schematic flowchart of an implementation process of Embodiment 2 of the present invention.
  • FIG. 4 is a schematic structural diagram of a third embodiment of the present invention.
  • FIG. 5 is a schematic structural diagram of a fourth embodiment of the present invention.
  • FIGS. 6(a)-6(c) are schematic diagrams showing an application to which an embodiment of the present invention is applied;
  • FIG. 7 is a schematic structural diagram of hardware components of an apparatus according to an embodiment of the present invention.
  • FIG. 1 is a schematic diagram of a hardware entity in which an image information recognition processing apparatus can be located in an embodiment of the present invention; as shown in FIG. 1, the image information recognition processing apparatus in the embodiment of the present invention can be located in a mobile phone 21, a notebook 22, an all-in-one 23, and a desktop computer.
  • the terminal such as 24, it may be located in the server 11 or in the base station. Of course, it can also be located in other terminals such as personal digital assistant PDA, e-reader, etc., and there are no excessive restrictions.
  • the terminal, server or base station can communicate with other terminals, servers or other base stations wirelessly or by wire.
  • FIG. 1 is only an example of an architecture for implementing an embodiment of the present invention, and the present invention is implemented.
  • the example is not limited to the structure described above with reference to Fig. 1, and various embodiments of the present invention are proposed based on the system architecture.
  • the first reference image is acquired first, and then a certain sub-image is obtained by performing a certain interception or cropping process on the image to be detected, and the first reference image and the sub-image are in color distribution information and/or image fingerprint information.
  • the image to be detected is an image similar to the first reference image.
  • the CF algorithm is used to obtain the color distribution information of the image, and the fingerprint information of the image is obtained by the phash algorithm, which can ensure that the image with high similarity can not be missed or the recognition accuracy can be improved.
  • FIG. 2 is a schematic flowchart of an implementation process according to Embodiment 1 of the present invention; as shown in FIG. 2, the method includes:
  • Step 201 Acquire a first reference image.
  • a picture with a high degree of similarity to the picture A itself or a part of the picture A may be found from other pictures, such as searching for a picture including an object such as a peony in the picture A in other pictures.
  • the first reference picture is the picture A itself.
  • the first reference picture is a partial image including the peony flower in the picture A.
  • the first predetermined image is first determined, and then the first predetermined image is intercepted or cropped according to the second preset rule to obtain the first reference image; the second preset rule is at least provided with the The position and size of the interception or cropping process of the first predetermined image.
  • the first predetermined image may be a local picture, a picture taken by a camera, a picture received from another terminal/server, or a picture downloaded from a network.
  • the first predetermined image may be the aforementioned picture A; the picture A may be completely intercepted or partially intercepted.
  • a picture A of 200*200 pixels it can be scaled by 1/5 (remove Off 40 pixels) Intercepting from the pixel (0,0) position, the captured picture is 160*160 pixels, including pixels (0,0) to (160,160); or, according to 1/ The ratio of 5 is intercepted from the pixel (5, 5) position, and the captured picture is 160*160 pixels, including pixels (5, 5) to (165, 165), depending on actual usage requirements.
  • Step 202 Acquire at least one image to be detected, perform an Nth interception or cropping process on the image to be detected in at least one specified direction according to a first preset rule, and obtain an Nth sub image in the at least one image to be detected. ;
  • the image to be detected may be a local picture, a picture taken by a camera, a picture received from another terminal/server, or a picture downloaded from a network.
  • the first preset rule is configured with at least a position and a size for performing an Nth interception or cropping process on the at least one image to be detected; the Nth sub image has the same size as the first reference image, N Is a positive integer.
  • the size of the image to be detected may be the same as or different from the size of the first predetermined image; the specified direction is the lateral direction and the longitudinal direction of the picture.
  • the interception in the lateral direction may be performed first, and the interception in the longitudinal direction is performed when the sub-image cut out from the lateral direction is similar to the first reference image.
  • the size of the Nth sub-image needs to be the same as the size of the first reference image.
  • the pixel is intercepted from the position of the pixel (0, 0), and the captured image, that is, the first sub-image is 160*160 pixels, including pixels (0, 0) to (160, 160);
  • the interception is performed from the position of the pixel (1, 0) according to the same 1/5 ratio.
  • the second sub-image after the interception is 160*160 pixels, including the pixel (1, 0). ⁇ (161, 160), and so on.
  • Step 203 Acquire a first attribute value and/or a second attribute value in the first reference image and the Nth sub-image, respectively, where the first attribute value is used to represent color distribution information of an image, where The second attribute value is used to represent the fingerprint information of the image;
  • the first attribute value is a color distribution histogram calculated by a CF algorithm
  • the second attribute value is fingerprint information of an image calculated by a phash algorithm
  • a color distribution histogram of the first reference image and the Nth sub-image is calculated.
  • image fingerprint information For details on the process of calculating the color distribution histogram and image fingerprint information, please refer to the existing related description, which is not described here.
  • Step 204 When the first attribute value and/or the second attribute value of the Nth sub-image and the corresponding attribute value of the first reference image satisfy a predetermined condition, identify that the Nth sub-image belongs to the The image to be detected matches the first reference image.
  • the similarity between the color distribution histogram of the Nth sub-image and the color distribution histogram of the first reference image is higher, and/or the image fingerprint of the Nth sub-image is higher in similarity to the image fingerprint of the first reference image.
  • the image to be detected to which the Nth sub-image belongs is an image similar to the first reference image.
  • the first reference image is acquired first, and then a certain sub-image is obtained by performing a certain interception or cropping process on the image to be detected, and the first reference image and the sub-image are in terms of color distribution information and/or image fingerprint information.
  • the image to be detected is an image similar to the first reference image.
  • the CF algorithm is used to obtain the color partial histogram
  • the phash algorithm is used to obtain the image fingerprint information, which can ensure that the high accuracy of the image can be ensured or improved.
  • FIG. 3 is a schematic flowchart of an implementation process of Embodiment 2 of the present invention; as shown in FIG. 3, the method includes:
  • Step 301 Acquire a first reference image.
  • a picture with a high degree of similarity to the picture A itself or a part of the picture A may be found from other pictures, such as searching for a picture including an object such as a peony in the picture A in other pictures.
  • the first reference picture is the picture A itself.
  • the first reference image is a partial image including a peony flower in the picture A.
  • the first predetermined image is first determined, and then the first predetermined image is intercepted or cropped according to the second preset rule to obtain the first reference image; the second preset rule is at least provided with the The position and size of the interception or cropping process of the first predetermined image.
  • the first predetermined image may be a local picture, a picture taken by a camera, a picture received from another terminal/server, or a picture downloaded from a network.
  • the first predetermined image may be the aforementioned picture A; the picture A may be completely intercepted or partially intercepted.
  • the interception can be performed from the position of the pixel (0, 0) according to the ratio of 1/5 (removing 40 pixels), and the picture after the interception is 160*160. Pixels, including pixels (0,0) to (160,160); or, at a ratio of 1/5, are taken from the pixel (5, 5) position, and the captured picture is 160*160 pixels, including Pixels (5, 5) to (165, 165) are determined according to actual usage.
  • Step 302 Acquire at least one image to be detected, perform at least one Nth interception or cropping process of the at least one specified direction on the image to be detected according to a first preset rule, to obtain an Nth sub image in the at least one image to be detected. ;
  • the image to be detected may be a local picture, a picture taken by a camera, a picture received from a terminal/server, or a picture downloaded from a network.
  • the first preset rule is configured with at least a position and a size for performing an Nth interception or cropping process on the at least one image to be detected; the Nth sub image has the same size as the first reference image, N Is a positive integer.
  • the size of the image to be detected may be the same as or different from the size of the first predetermined image.
  • the specified direction is a lateral direction and a longitudinal direction of the picture.
  • the intercepting in the lateral direction may be performed first, and when the sub-image taken from the lateral direction does not exist similarly to the first reference image, the longitudinal direction is further performed. Intercept.
  • the size of the Nth sub-image needs to be the same as the size of the first reference image.
  • picture B is 200*200 pixels, intercepted in the horizontal direction, and panned each time.
  • the picture is the first sub-image is 160 * 160 pixels, including pixels (0,0) ⁇ (160, 160); in the second interception according to the ratio of 1 / 5, from the pixel point (1, 0) position At the beginning, the interception is performed, and the second sub-image is 160*160 pixels, including pixels (1, 0) to (161, 160), and so on.
  • Step 303 Acquire a first attribute value and/or a second attribute value in the first reference image and the Nth sub-image, respectively, where the first attribute value is used to represent color distribution information of an image, where The second attribute value is used to represent the fingerprint information of the image;
  • the first attribute value is a color distribution histogram calculated by a CF algorithm
  • the second attribute value is fingerprint information of an image calculated by a phash algorithm
  • a color distribution histogram of the first reference image and the Nth sub-image is calculated.
  • image fingerprint information For details on the process of calculating the color distribution histogram and image fingerprint information, please refer to the existing related description, which is not described here.
  • Step 304 Determine first data according to the first reference image and the first attribute value of the Nth sub-image, where the first data is a degree of similarity between the first reference image and the Nth sub-image.
  • the similarity value of the first reference image and the Nth sub-image with respect to the color distribution histogram is calculated by the cosine similarity algorithm, the first data is obtained, and the first data is buffered.
  • the first data can also be obtained by other algorithms.
  • Step 305 Acquire second data, where the second data is characterized as a maximum value in the first data of each of the N-1 sub-images before the Nth sub-image;
  • the first N-1 sub-images of the image to be detected are acquired to the first N-1 sub-images
  • the similarity values of each of the first N-1 sub-images and the first reference image are calculated, Cache it up for use in subsequent scenarios.
  • the first data corresponding to the previous N-1 sub-images that have been cached is read, and the maximum value is calculated as the second data.
  • Step 306 Determine, according to the first data, the second data, the second attribute value of the first reference image, and the second attribute value of the Nth sub-image, a first attribute value of the Nth sub-image and/or Or whether the second attribute value and the corresponding attribute value of the first reference image satisfy a predetermined condition;
  • the at least one image to be detected is subjected to a clipping or cropping process in a second specified direction, and the Nth sub-image is performed on the at least one image to be detected
  • the interception or cropping process in the first specified direction that is, the interception direction of the image to be measured is changed.
  • the first data of the Nth sub-image is not the maximum value, indicating that the Nth sub-image does not have the possibility of being similar to the first reference image, and changing the interception direction of the image to be detected.
  • the interception position and the ratio may be the same as when the lateral direction is intercepted, obtaining a corresponding sub-image, and determining whether the sub-image is similar to the first reference image, the process and the foregoing The content is similar and will not be described.
  • the first data of the Nth sub-image is a maximum value, determining that the first attribute value of the Nth sub-image and the first attribute value of the first reference image satisfy a predetermined condition, and it is considered that the image to be detected is intercepted in the lateral direction.
  • the Nth sub-image is likely to be similar to the first reference image or the image to be detected is likely to be similar to the first reference image; then the difference between the Nth sub-image and the first reference image on the second attribute value is calculated;
  • the difference between the second attribute value of the Nth sub-image and the second attribute value of the first reference image does not exceed the first predetermined range, such as the Hamming distance of the image fingerprint information of the Nth sub-image and the first reference image is less than or equal to M.
  • Step 307 When the first attribute value and/or the second attribute value of the Nth sub-image and the corresponding attribute value of the first reference image satisfy a predetermined condition, identify that the Nth sub-image belongs to the The image to be detected matches the first reference image.
  • the first attribute value of the N-th sub-image is considered to be And/or the second attribute value and the corresponding attribute value of the first reference image satisfy a predetermined condition, and at this time, it can be confirmed that the image to be detected belonging to the Nth sub-image is matched with the first reference image Similar pictures.
  • the method further includes:
  • the Nth interception or cropping process in the specified direction performed on the at least one image to be detected is not the last interception or cropping process in the first specified direction, performing the at least one image to be detected Obtaining an N+1th interception or cropping process in the specified direction to obtain an N+1th sub-image; acquiring a first attribute value and/or a second attribute value in the (N+1)th sub-image; When the first attribute value and/or the second attribute value of the N+1 sub-images and the corresponding attribute value of the first reference image satisfy a predetermined condition, the image to be detected to which the (N+1)th sub-image belongs is identified Matching the first reference image.
  • the N+1th interception in the specified direction is performed on the image to be detected, and the N+1th sub-image is obtained, and the CF algorithm is used to obtain the first Color distribution histogram of N+1 sub-images and/or image fingerprint obtained by phash algorithm, and then according to color distribution histogram And the image fingerprint, determining whether the N+1th sub-image or the image to be detected belongs to the first reference image.
  • the N+1th interception in the specified direction such as the horizontal direction
  • the CF algorithm is used to obtain the first Color distribution histogram of N+1 sub-images and/or image fingerprint obtained by phash algorithm, and then according to color distribution histogram And the image fingerprint, determining whether the N+1th sub-image or the image to be detected belongs to the first reference image.
  • the Nth interception or cropping process in the specified direction performed on the at least one image to be detected is the last interception or cropping process in the first specified direction, performing the second image on the at least one image to be detected Intercept or crop processing in the specified direction.
  • the interception of the image to be detected first in the lateral direction first designated direction
  • the interception is performed one by one in a certain ratio in the lateral direction, and it is not determined that there is a sub-image similar to the first reference image
  • a predetermined direction such as the longitudinal direction (the second specified direction) according to the predetermined ratio and the interception position
  • Changing the first reference image can be changed by continuously intercepting the first reference image.
  • the first reference image selected for the first time is the pixel point (5, 5) to (165, 165) of the picture A
  • the first reference image selected for the second time may be the pixel point of the picture A (10, 5) ⁇ (170, 165)
  • This part, the second selection of this part of the image can be obtained by shifting the first reference image by 5 pixels in the lateral direction.
  • the subsequent processing of capturing the sub-image in the vertical direction is similar to the subsequent processing of the sub-image obtained by intercepting in the lateral direction, and details are not described herein again.
  • the color histogram is first adopted.
  • the degree of similarity of the figure determines whether the Nth sub-image has the possibility of being similar to the first reference image. If it is determined by the degree of similarity of the color histogram that the Nth sub-image has the possibility of being similar to the first reference image, It is determined by the element of the image fingerprint information whether the Nth sub-image has a similarity with the first reference image. If there is a similar possibility by the two elements mentioned above, the Nth sub-image or its affiliated component is considered to be detected. The image is similar to the first reference image. The combination of the two algorithms ensures that the image with high similarity is not missed or the recognition accuracy is improved.
  • the first reference image is acquired first, and then a certain sub-image is obtained by performing a certain interception or cropping process on the image to be detected, and the first reference image and the sub-image are in terms of color distribution information and/or image fingerprint information.
  • the image to be detected is an image similar to the first reference image.
  • the first reference image and the sub-image are similar in terms of color distribution information and image fingerprint information
  • the image to be detected is considered to be an image similar to the first reference image.
  • the CF algorithm is used to obtain the color partial histogram
  • the phash algorithm is used to obtain the image fingerprint information. The combination of the two algorithms ensures that the high similarity image can be ensured or improved.
  • FIG. 4 is a schematic structural diagram of a first embodiment of an apparatus according to an embodiment of the present invention; as shown in FIG. 4, the apparatus includes: a first obtaining unit 401, a second obtaining unit 402, a third obtaining unit 403, and a first identifier. Unit 404; wherein
  • the first obtaining unit 401 is configured to acquire a first reference image
  • a picture with a high degree of similarity to the picture A itself or a part of the picture A may be found from other pictures, such as searching for a picture including an object such as a peony in the picture A in other pictures.
  • the first reference picture is the picture A itself.
  • the first reference picture is a partial image including the peony flower in the picture A.
  • the first acquiring unit 401 first determines a first predetermined image, and then performs intercepting or cropping processing on the first predetermined image according to the second preset rule to obtain the first reference image; and the second preset rule is configured at least The position and size of the interception or cropping process of the first predetermined image.
  • the first predetermined image may be a local picture, a picture taken by a camera, a picture received from another terminal/server, or a picture downloaded from a network.
  • the first predetermined image may be the aforementioned picture A; the picture A may be completely intercepted, or may be used, or For partial interception.
  • the interception can be performed from the position of the pixel (0, 0) according to the ratio of 1/5 (removing 40 pixels), and the picture after the interception is 160*160. Pixels, including pixels (0,0) to (160,160); or, at a ratio of 1/5, are taken from the pixel (5, 5) position, and the captured picture is 160*160 pixels, including Pixels (5, 5) ⁇ (165, 165).
  • the second obtaining unit 402 is configured to acquire at least one image to be detected, and perform an Nth interception or cropping process on the image to be detected in at least one specified direction according to the first preset rule, to obtain the at least one image to be detected.
  • the Nth sub-image, N is a positive integer;
  • the image to be detected may be a local picture, a picture taken by a camera, a picture received from another terminal/server, or a picture downloaded from a network.
  • the first preset rule is configured with at least a position and a size for performing an Nth interception or cropping process on the at least one image to be detected; the Nth sub image has the same size as the first reference image, N Is a positive integer.
  • the size of the image to be detected may be the same as or different from the size of the first predetermined image; the specified direction is the lateral direction and the longitudinal direction of the picture, and the preferred second acquiring unit 402 may first perform the interception in the lateral direction, from the lateral direction.
  • the sub-image taken up is similar to the first reference image, and the interception in the longitudinal direction is performed.
  • the size of the Nth sub-image needs to be the same as the size of the first reference image.
  • the picture B is 200*200 pixels, and one pixel point of the picture is panned in the horizontal direction, and the second acquisition unit 402 performs the first interception according to the ratio of 1/5 (excluding 200*1).
  • the first sub-image is 160*160 pixels, including pixels (0,0) to (160, 160); when the second acquisition unit 402 performs the second interception, the interception is performed from the position of the pixel point (1, 0) according to the ratio of 1/5, and the second sub-image after the interception is 160*160 pixels. It includes pixels (1,0) to (161,160), and so on.
  • the third obtaining unit 403 is configured to separately acquire the first reference image and the Nth a first attribute value and/or a second attribute value in the sub-image, the first attribute value being used to characterize color distribution information of the image, the second attribute value being used to represent fingerprint information of the image;
  • the first attribute value is a color distribution histogram calculated by a CF algorithm
  • the second attribute value is fingerprint information of an image calculated by a phash algorithm
  • the third obtaining unit 403 calculates a first reference image and an Nth sub-image.
  • the color distribution histogram and/or image fingerprint information For details on the process of calculating the color distribution histogram and image fingerprint information, please refer to the existing related description, which is not described here.
  • the first identifying unit 404 is configured to identify the Nth child when the first attribute value and/or the second attribute value of the Nth sub-image and the corresponding attribute value of the first reference image satisfy a predetermined condition The image to be detected to which the image belongs is matched with the first reference image.
  • the first identifying unit 404 confirms that the color distribution histogram of the Nth sub image has a higher degree of similarity with the color distribution histogram of the first reference image, and/or the image fingerprint of the Nth sub image and the first reference image The image fingerprint similarity is high, and it is confirmed that the image to be detected to which the Nth sub-image belongs is an image similar to the first reference image.
  • the first reference image is acquired first, and then a certain sub-image is obtained by performing a certain interception or cropping process on the image to be detected, and the first reference image and the sub-image are in terms of color distribution information and/or image fingerprint information.
  • the image to be detected is an image similar to the first reference image.
  • the CF algorithm is used to obtain the color partial histogram
  • the phash algorithm is used to obtain the image fingerprint information, which can ensure that the high accuracy of the image can be ensured or improved.
  • FIG. 5 is a schematic structural diagram of a second embodiment of an apparatus according to an embodiment of the present invention. As shown in FIG. 5, the apparatus includes: a first obtaining unit 501, a second obtaining unit 502, a third obtaining unit 503, and a first identifier. Unit 504; wherein
  • the first obtaining unit 501 is configured to acquire a first reference image
  • a picture with a high degree of similarity to the picture A itself or a part of the picture A may be found from other pictures, such as searching for a picture including an object such as a peony in the picture A in other pictures.
  • the first reference picture is the picture A itself.
  • the first reference picture is a partial image including the peony flower in the picture A.
  • the first acquiring unit 501 first determines a first predetermined image, and then performs intercepting or cropping processing on the first predetermined image according to the second preset rule to obtain the first reference image; and the second preset rule is configured at least with The position and size of the interception or cropping process of the first predetermined image.
  • the first predetermined image may be a local picture, a picture taken by a camera, a picture received from another terminal/server, or a picture downloaded from a network.
  • the first predetermined image may be the aforementioned picture A; the picture A may be completely intercepted or partially intercepted.
  • the interception can be performed from the position of the pixel (0, 0) according to the ratio of 1/5 (removing 40 pixels), and the picture after the interception is 160*160. Pixels, including pixels (0,0) to (160,160); or, at a ratio of 1/5, are taken from the pixel (5, 5) position, and the captured picture is 160*160 pixels, including Pixels (5, 5) ⁇ (165, 165).
  • the second obtaining unit 502 is configured to acquire at least one image to be detected, and perform an Nth interception or cropping process on the image to be detected in at least one specified direction according to the first preset rule, to obtain the at least one image to be detected.
  • the Nth sub-image, N is a positive integer;
  • the image to be detected may be a local picture, a picture taken by a camera, a picture received from another terminal/server, or a picture downloaded from a network.
  • the first preset rule is configured with at least a position and a size for performing an Nth interception or cropping process on the at least one image to be detected; the Nth sub image has the same size as the first reference image, N Is a positive integer.
  • the size of the image to be detected may be the same as or different from the size of the first predetermined image;
  • the specified direction is the horizontal direction and the vertical direction of the picture.
  • the second obtaining unit 502 may first perform the intercepting in the lateral direction, and perform the intercepting in the longitudinal direction when the sub-image taken from the lateral direction is similar to the first reference image.
  • the size of the Nth sub-image needs to be the same as the size of the first reference image.
  • the third acquiring unit 503 is configured to respectively acquire a first attribute value and/or a second attribute value in the first reference image and the Nth sub-image, where the first attribute value is used to represent a color distribution of the image Information, the second attribute value is used to represent fingerprint information of the image;
  • the first attribute value is a color distribution histogram calculated by a CF algorithm
  • the second attribute value is fingerprint information of an image calculated by a phash algorithm
  • the third obtaining unit 503 calculates a first reference image and an Nth sub-image.
  • the color distribution histogram and/or image fingerprint information For details on the process of calculating the color distribution histogram and image fingerprint information, please refer to the existing related description, which is not described here.
  • the third obtaining unit 503 is further configured to determine, according to the first reference image and the first attribute value of the Nth sub-image, the first data, where the first data is the first reference image and the Nth sub-image a degree of similarity in color; acquiring second data, the second data being a maximum value in the first data of each of the N-1 sub-images preceding the Nth sub-image;
  • the third obtaining unit 503 calculates the first reference image and the Nth sub-image about the color by the cosine similarity algorithm according to the color distribution histogram of the first reference image and the Nth sub-image.
  • the similarity value of the color distribution histogram is cached in the cache unit (not shown in Figure 5). Of course, it can also be obtained by other algorithms.
  • the third acquisition unit 503 calculates the sub-images of the first N-1 sub-images and the first reference image.
  • the similarity value is cached in the cache unit for use in subsequent scenarios.
  • the third obtaining unit 503 reads the first data corresponding to the first N-1 sub-images previously buffered in the buffer unit, and calculates the maximum value thereof as the second data.
  • the first identifying unit 504 is configured to determine the Nth sub image according to the first data, the second data, the second attribute value of the first reference image, and the second attribute value of the Nth sub image. Whether the first attribute value and/or the second attribute value meets a predetermined condition with a corresponding attribute value of the first reference image.
  • the first identifying unit 504 determines, in the first data of the second data and the Nth sub-image, whether the first data of the Nth sub-image is a maximum value
  • the first data of the Nth sub-image is not the maximum value, triggering the second obtaining unit 502 to perform the intercepting or cropping processing in the second specified direction on the at least one to-be-detected image, where the Nth sub-image is The intercepting or cropping processing in the first specified direction by the at least one image to be detected, that is, changing the intercepting direction of the image to be tested.
  • the first data of the Nth sub-image is not the maximum value, indicating that the Nth sub-image does not have the possibility of being similar to the first reference image, and changing the interception direction of the image to be detected.
  • the first identification unit 504 determines that the first data of the Nth sub-image is a maximum value, determining that the first attribute value of the Nth sub-image and the first attribute value of the first reference image satisfy a predetermined condition, and Intercepting the image to be detected in the lateral direction to obtain the Nth sub-image is likely to be similar to the first reference image or the image to be detected is likely to be similar to the first reference image;
  • the first identifying unit 504 calculates a difference between the Nth sub-image and the first reference image on the second attribute value; when the difference between the second attribute value of the Nth sub-image and the second attribute value of the first reference image Determining a Hamming distance of the image fingerprint information of the Nth sub-image and the first reference image that is less than or equal to M, determining a second attribute value of the Nth sub-image and a second of the first reference image The attribute value satisfies the predetermined condition.
  • the first identifying unit 504 calculates that the difference between the second attribute value of the Nth sub-image and the second attribute value of the first reference image exceeds the first predetermined range, such as the image of the Nth sub-image and the first reference image
  • the Hamming distance of the fingerprint information is greater than M
  • the first identifying unit 504 is configured to identify the Nth child when the first attribute value and/or the second attribute value of the Nth sub-image and the corresponding attribute value of the first reference image satisfy a predetermined condition The image to be detected to which the image belongs is matched with the first reference image.
  • the first identifying unit 504 determines that the first attribute value and/or the second attribute value of the Nth sub-image and the corresponding attribute value of the first reference image do not satisfy the predetermined condition
  • the first identifying unit 504 determines whether the Nth interception or cropping process in the specified direction performed on the at least one image to be detected is the last intercepting or cropping process in the specified direction;
  • the first identification unit 504 determines that the Nth interception or cropping process in the specified direction performed on the at least one image to be detected is not the last interception or cropping process in the first specified direction, triggering the second
  • the obtaining unit 502 performs an (N+1)th interception or cropping process on the at least one image to be detected in the specified direction to obtain an N+1th sub-image; and correspondingly, the third obtaining unit 503 acquires the N+th a first attribute value and/or a second attribute value in one sub-image; the first identification unit 504 confirms that the first attribute value of the (N+1)th sub-image and/or the second When the attribute value and the corresponding attribute value of the first reference image satisfy a predetermined condition, it is recognized that the image to be detected to which the N+1th sub-image belongs is matched with the first reference image.
  • the second obtaining unit 502 performs the N+1th interception in the specified direction, such as the horizontal direction, to obtain the N+1th sub-image
  • the third obtaining unit 503 obtains the N+1th through the CF algorithm.
  • the color distribution histogram of the sub-images and/or the image fingerprints are obtained by the phash algorithm
  • the first identifying unit 504 determines, according to the color distribution histogram and the image fingerprints, whether the N+1th sub-image or the image to be detected thereof is similar to the first image.
  • a reference image please refer to the foregoing description of the Nth sub-image for the specific process.
  • the first identification unit 504 determines that the Nth interception or cropping process in the specified direction performed on the at least one image to be detected is the last interception or cropping process in the first specified direction, triggering the second acquisition
  • the unit 502 performs the intercepting or cropping process in the second specified direction on the at least one image to be detected.
  • taking the first specified direction of the image to be detected as the first direction for example, the horizontal direction is taken as an example.
  • the first identifying unit 504 When the second acquiring unit 502 is intercepted one by one according to a certain ratio in the lateral direction, the first identifying unit 504 also does not determine the first If the sub-images of the reference image are similar, the second acquisition unit 502 triggers the capture in the second specified direction, such as the longitudinal direction, according to a predetermined ratio and the interception position. The first recognition unit 504 compares whether the sub-images taken in the longitudinal direction are compared with the sub-images. The first reference images are similar, and if they are still not similar, triggering the first switching unit (not illustrated in Figure 5) to change the first reference image can be changed by continuously intercepting the first reference image.
  • the first reference image selected for the first time is the pixel point (5, 5) to (165, 165) of the picture A
  • the first reference image selected for the second time may be the pixel point of the picture A (10, 5) ⁇ (170, 165)
  • This part, the second selection of this part of the image can be obtained by shifting the first reference image by 5 pixels in the lateral direction.
  • the subsequent processing of capturing the sub-image in the vertical direction is similar to the subsequent processing of the sub-image obtained by intercepting in the lateral direction, and details are not described herein again.
  • first determining whether the Nth sub-image exists similar to the first reference image by using the similarity degree of the color histogram It is possible to determine, by the similarity degree of the color histogram, that the Nth sub-image has a possibility similar to the first reference image, and then determine whether the Nth sub-image exists similarly to the first reference image by using the image fingerprint information element.
  • the first reference image is acquired first, and then a certain sub-image is obtained by performing a certain interception or cropping process on the image to be detected, and the first reference image and the sub-image are in terms of color distribution information and/or image fingerprint information.
  • the image to be detected is an image similar to the first reference image.
  • the first reference image and the sub-image are similar in terms of color distribution information and image fingerprint information
  • the image to be detected is considered to be an image similar to the first reference image.
  • the CF algorithm is used to obtain the color partial histogram
  • the phash algorithm is used to obtain the image fingerprint information. The combination of the two algorithms ensures that the high similarity image can be ensured or improved.
  • Step 700 Acquire a first reference image
  • the terminal supports full-screen display of picture A (first predetermined image) having a pixel of 200*200, and picture A includes user 1, in proportion to 1/5 (40 pixels are removed) Point) is taken from the position of the pixel (0, 0) (ie, the upper left corner of the picture), and the intercepted picture, that is, the first reference image is 160*160 pixels, including pixels (0, 0) to (160, 160) (as shown by the black gray box in Figure 7(a)).
  • Step 701 Acquire at least one image to be detected.
  • the image to be detected is a picture B as shown in FIG. 6(b).
  • Step 702 Perform an Nth interception or cropping process in the horizontal direction on the to-be-detected image according to the first preset rule to obtain a current sub-image (the Nth sub-image);
  • the image to be detected is a picture B (including one hand) as shown in FIG. 6(b), and it is assumed that the picture B itself is also 200*200 pixels, and is also horizontally at a ratio of 1/5.
  • the direction is intercepted, and each interception is shifted by one pixel in the positive direction of the horizontal direction than the previous time.
  • the first time from the pixel point (0, 0) the 1/5 ratio is intercepted, and the second time is from (1, 0)
  • the 1/5 ratio is cut at the pixel, and the 1/5 ratio is taken from the (2,0) pixel at the 3rd time.
  • the obtained fourth sub-image includes the pixel points (4, 0) to (164, 160) of the picture B, as shown by the black gray box in FIG. 6(b).
  • Step 703 a color distribution histogram of the first reference image and the current sub-image calculated by the CF algorithm; and calculating image fingerprint information of the first reference image and the current sub-image by using a phash algorithm;
  • Step 704 Calculate a similarity value of the first reference image and the current sub-image on the color distribution according to the first reference image and the color distribution histogram of the current sub-image; and read the second data;
  • Step 705 Determine a size of the second data and the first data.
  • step 706 is performed.
  • step 709 is performed.
  • Step 706 Calculate a difference between the current sub-image and the first reference image on the image fingerprint information
  • step 708 If the Hamming distance of the image fingerprint information of the Nth sub-image and the first reference image is greater than M, it is considered that the difference between the two images is larger on the element of the image fingerprint information, and step 708 is performed;
  • Step 707 the picture B to which the current sub-image belongs is similar to the first reference image, and the process ends;
  • the Nth sub-image may also be considered to be an image similar to the first reference image.
  • Step 708 Determine whether the Nth interception or cropping process in the lateral direction of the at least one image to be detected is the last interception or cropping process in the lateral direction;
  • step 709 If the determination is yes, go to step 709;
  • step 710 is performed
  • Step 709 Performing a cropping or cropping process on the at least one image to be detected in the longitudinal direction to obtain a sub-image captured in the longitudinal direction as the current sub-image and returning to step 703;
  • Step 710 Perform an N+1th interception or cropping process on the at least one image to be detected in the horizontal direction to obtain an N+1th sub-image as the current sub-image and return to step 703.
  • the first reference image is intercepted or cropped from the first predetermined image, and the Nth sub-image is obtained by performing the Nth interception or cropping from the image to be detected, if the Nth sub-image There is a high degree of similarity with the first reference image on the color distribution histogram, and it is confirmed from the image fingerprint information whether the Nth sub-image and the first reference image confirm similarity. If there is a similarity in both the color distribution and the image fingerprint, the image to which the Nth sub-image belongs is considered to be an image in which the image to be detected is similar to the first reference image.
  • the CF algorithm is used to obtain the color distribution histogram
  • the phash algorithm is used to obtain the image fingerprint.
  • the two algorithms are combined and the CF algorithm is used as the preliminary screening algorithm (the preliminary screening may be similar to the first reference image).
  • the picture does not miss the picture with high similarity.
  • the phash algorithm can be used as a method of further confirmation to ensure or improve the recognition accuracy.
  • An image information recognition processing device that is integrated into one or each unit function split setting for realizing each unit function includes at least a database for storing data and a processor for data processing, or includes a server disposed in the server Storage media or storage media that are set independently.
  • a microprocessor for the processor for data processing, a microprocessor, a central processing unit (CPU), a digital signal processor (DSP, Digital Singnal Processor) or programmable logic may be used when performing processing.
  • An FPGA Field-Programmable Gate Array
  • an operation instruction for a storage medium, an operation instruction, which may be a computer executable code, by which the image information recognition processing method of the embodiment of the present invention is implemented Each step in the process.
  • the apparatus includes a processor 31, a storage medium 32, and at least one external communication interface 33; the processor 31, the storage medium 32, and the external communication interface 33 are all connected by a bus 34.
  • the embodiment of the present invention further provides a computer storage medium, wherein computer executable instructions are stored, and the computer executable instructions are configured to execute the image information recognition processing method.
  • the integrated modules described in the embodiments of the present invention may also be stored in a computer readable storage medium if they are implemented in the form of software functional modules and sold or used as separate products. Based on such understanding, those skilled in the art will appreciate that embodiments of the present application can be provided as A method, system, or computer program product. Thus, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment in combination of software and hardware. Moreover, the application can take the form of a computer program product embodied on one or more computer-usable storage media containing computer usable program code, including but not limited to a USB flash drive, a mobile hard drive, a read only memory ( ROM, Read-Only Memory), 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.
  • the similarity between the first attribute value and the second attribute value is obtained from the Nth sub-image and the first reference image, that is, both the color distribution of the image and the image fingerprint information are taken into consideration.
  • the effect of image similarity recognition; using the embodiment of the invention can effectively improve the recognition accuracy rate and reduce the probability that similar pictures are missed.

Abstract

一种图像信息识别处理方法、计算机存储介质，所述方法包括：获取第一基准图像（201）；获取至少一个待检测图像，按照第一预设规则对所述待检测图像进行至少一个指定方向的第N次截取或裁剪处理，得到所述至少一个待检测图像中的第N个子图像，N为正整数（202）；分别获取所述第一基准图像和所述第N个子图像中的第一属性值和/或第二属性值，所述第一属性值用于表征图像的颜色分布信息，所述第二属性值用于表征图像的指纹信息（203）；当所述第N个子图像的第一属性值和/或第二属性值与所述第一基准图像的对应属性值满足预定条件时，识别出所述第N个子图像隶属的所述待检测图像与所述第一基准图像相匹配（204）。

Description

一种图像信息识别处理方法及装置、计算机存储介质 技术领域
本发明涉及图像信息处理技术，具体涉及一种图像信息识别处理方法及装置、计算机存储介质。
背景技术
在图像处理领域，相似图像的识别技术越发成熟。目前，常用的图像相似识别方法包括：颜色分布直方图CF算法及感知哈希phash算法。其中，CF算法存在有很大的错误识别概率，即很容易将与基准图像不相似的图像错误的认为相似；phash算法在一定程度上虽然可保证识别正确率，但是其容易漏掉一些相似度较高的图片。可见，亟需一种既能够保证不漏掉相似度高的图片也可以保证或提高识别正确率的解决方案。
发明内容
为解决现有存在的技术问题，本发明实施例提供一种图像信息识别处理方法及装置、计算机存储介质，以至少解决现有技术存在的问题，提高识别正确率、降低相似图片被漏掉的概率。
本发明实施例的技术方案是这样实现的：
本发明实施例提供了一种图像信息识别处理方法，所述方法包括：
获取第一基准图像；
获取至少一个待检测图像，按照第一预设规则对所述待检测图像进行至少一个指定方向的第N次截取或裁剪处理，得到所述至少一个待检测图像中的第N个子图像，N为正整数；
分别获取所述第一基准图像和所述第N个子图像中的第一属性值和/或 第二属性值，所述第一属性值用于表征图像的颜色分布信息，所述第二属性值用于表征图像的指纹信息；
当所述第N个子图像的第一属性值和/或第二属性值与所述第一基准图像的对应属性值满足预定条件时，识别出所述第N个子图像隶属的所述待检测图像与所述第一基准图像相匹配。
本发明实施例提供一种图像信息识别处理装置，所述装置包括：
第一获取单元，配置为获取第一基准图像；
第二获取单元，配置为获取至少一个待检测图像，按照第一预设规则对所述待检测图像进行至少一个指定方向的第N次截取或裁剪处理，得到所述至少一个待检测图像中的第N个子图像，N为正整数；
第三获取单元，配置为分别获取所述第一基准图像和所述第N个子图像中的第一属性值和/或第二属性值，所述第一属性值用于表征图像的颜色分布信息，所述第二属性值用于表征图像的指纹信息；
第一识别单元，配置为当所述第N个子图像的第一属性值和/或第二属性值与所述第一基准图像的对应属性值满足预定条件时，识别出所述第N个子图像隶属的所述待检测图像与所述第一基准图像相匹配。
本发明实施例还提供一种计算机存储介质，其中存储有计算机可执行指令，该计算机可执行指令配置执行上述图像信息识别处理方法。
本发明实施例提供的图像信息识别处理方法，获取第一基准图像；获取至少一个待检测图像，按照第一预设规则对所述待检测图像进行至少一个指定方向的第N次截取或裁剪处理，得到所述至少一个待检测图像中的第N个子图像，N为正整数；分别获取所述第一基准图像和所述第N个子图像中图像的颜色分布信息和图像的指纹信息；当所述第N个子图像的颜色分布信息和/或图像的指纹信息与所述第一基准图像的对应信息满足预定条件时，确定所述第N个子图像隶属的所述待检测图像与所述第一基准图 像相匹配。
采用本发明实施例，从第N个子图像和第一基准图像在第一属性值和第二属性值等方面的相似性入手，也就是兼顾到图像的颜色分布情况和图像指纹信息这两个方面对图像相似识别的影响；采用本发明实施例可有效提高识别正确率，减少相似图片被漏掉的概率。
附图说明
图1为本发明实施例图像信息识别处理装置可位于的硬件实体的示意图；
图2为本发明实施例一的一个实现流程示意图；
图3为本发明实施例二的一个实现流程示意图；
图4为本发明实施例三的组成结构示意图；
图5为本发明实施例四的组成结构示意图；
图6(a)～6(c)为应用本发明实施例的一应用的示意图；
图7为本发明实施例中的一个装置的硬件组成结构示意图。
具体实施方式
以下结合附图对本发明的优选实施例进行详细说明，应当理解，以下所说明的优选实施例仅用于说明和解释本发明，并不用于限定本发明。
图1为本发明实施例图像信息识别处理装置可位于的硬件实体的示意图；如图1所示，本发明实施例的图像信息识别处理装置可以位于手机21、笔记本22、一体机23、台式机24等终端中，可以位于服务器11，也可以位于基站中。当然，还可以位于其它终端如个人数字助理PDA、电子阅读器中等，此处不做过多的限制。此外，终端、服务器或基站可通过无线方式或有线方式与其他终端、服务器或其他基站进行通信。
上述图1的例子只是实现本发明实施例的一个架构实例，本发明实施 例并不限于上述图1所述的结构，基于该系统架构，提出本发明各个实施例。
本发明实施例中，先获取第一基准图像，再通过对待检测图像进行某次截取或裁剪处理得到一个子图像，当第一基准图像与该子图像在颜色分布信息方面和/或图像指纹信息方面存在有相似时，认为待检测图像是与第一基准图像相似的图像。其中，利用了CF算法得到图像的颜色分布信息，利用phash算法得到了图像的指纹信息，可保证不漏掉相似度高的图片也可以保证或提高识别正确率。
实施例一
图2为本发明实施例一的一个实现流程示意图；如图2所示，所述方法包括：
步骤201：获取第一基准图像；
在实际应用中，可以从其它图片中查找到与图片A本身或图片A的部分图像相似度高的图片，如在其它图片中查找包括有图片A中某个对象如牡丹花的图片。当想要从其它图片中查找与图片A本身相似度高的图片时，第一基准图像为图片A本身。当想要从其它图片中查找与图片A的部分图像如牡丹花图像相似度高的图片时，第一基准图像为图片A中包括有牡丹花的部分图像。
本步骤中，先确定第一预定图像，然后按照第二预设规则对第一预定图像进行截取或裁剪处理，得到所述第一基准图像；所述第二预设规则至少设置有对所述第一预定图像的截取或裁剪处理的位置及大小。所述第一预定图像可以为本地图片、通过摄像头拍摄而得的图片、从其它终端/服务器中接收到的图片或从网络下载的图片。例如，第一预定图像可以为前述的图片A；应使用需求，可以对图片A进行全部截取，也可以进行部分截取。例如，对于200*200像素的图片A来说，可以按照1/5的比例(去除 掉40个像素点)从像素点(0，0)位置处开始进行截取，截取后的图片为160*160像素，包括像素点(0，0)～(160，160)；或者，按照1/5的比例从像素点(5，5)位置处开始进行截取，截取后的图片为160*160像素，包括像素点(5，5)～(165，165)，根据实际使用需求而定。
步骤202：获取至少一个待检测图像，按照第一预设规则对所述待检测图像进行至少一个指定方向的第N次截取或裁剪处理，得到所述至少一个待检测图像中的第N个子图像；
这里，所述待检测图像可以为本地图片、通过摄像头拍摄而得的图片、从其它终端/服务器中接收到的图片或从网络下载的图片。所述第一预设规则至少设置有对所述至少一个待检测图像进行第N次截取或裁剪处理的位置及大小；所述第N个子图像与所述第一基准图像具有相同的大小，N为正整数。待检测图像的大小可以与第一预定图像的大小相同或不同；所述指定方向为图片的横向方向和纵向方向。优选的，可以先进行横向方向上的截取，通过从横向方向上截取下的子图像与第一基准图像不存在相似可能时再进行纵向方向上的截取。第N个子图像的大小需要与第一基准图像的大小相同。例如，图片B为200*200个像素点，以横向方向、每次平移图片的一个像素点为例，进行第1次截取时按照1/5的比例(去除掉200*1/5＝40个像素点)从像素点(0，0)位置处开始进行截取，截取后的图片即第1个子图像为160*160像素，包括有像素点(0，0)～(160，160)；进行第2次截取时按照同样的1/5比例，从像素点(1，0)位置处开始进行截取，截取后的图片即第2个子图像为160*160像素，包括有像素点(1，0)～(161，160)，以此类推。
步骤203：分别获取所述第一基准图像和所述第N个子图像中的第一属性值和/或第二属性值，所述第一属性值用于表征图像的颜色分布信息，所述第二属性值用于表征图像的指纹信息；
这里，所述第一属性值为通过CF算法计算出的颜色分布直方图，第二属性值通过phash算法计算出的图像的指纹信息；计算第一基准图像、第N个子图像的颜色分布直方图和/或图像指纹信息。具体的计算颜色分布直方图、图像指纹信息的过程请参见现有相关说明，此处不赘述。
步骤204：当所述第N个子图像的第一属性值和/或第二属性值与所述第一基准图像的对应属性值满足预定条件时，识别出所述第N个子图像隶属的所述待检测图像与所述第一基准图像相匹配。
这里，当第N个子图像的颜色分布直方图与第一基准图像的颜色分布直方图的相似度较高、和/或第N个子图像的图像指纹与第一基准图像的图像指纹相似度较高时，确认第N个子图像隶属的待检测图像为与第一基准图像相似的图像。
本发明实施例中，先获取第一基准图像，再通过对待检测图像进行某次截取或裁剪处理得到一个子图像，当第一基准图像与子图像在颜色分布信息方面和/或图像指纹信息方面存在有相似时，认为待检测图像是与第一基准图像相似的图像。其中，利用CF算法得到颜色部分直方图，利用phash算法得到图像指纹信息，可保证不漏掉相似度高的图片也可以保证或提高识别正确率。
实施例二
图3为本发明实施例二的一个实现流程示意图；如图3所示，所述方法包括：
步骤301：获取第一基准图像；
在实际应用中，可以从其它图片中查找到与图片A本身或图片A的部分图像相似度高的图片，如在其它图片中查找包括有图片A中某个对象如牡丹花的图片。当想要从其它图片中查找与图片A本身相似度高的图片时，第一基准图像为图片A本身。当想要从其它图片中查找与图片A的部分图 像如牡丹花图像相似度高的图片时，第一基准图像为图片A中包括有牡丹花的部分图像。
本步骤中，先确定第一预定图像，然后按照第二预设规则对第一预定图像进行截取或裁剪处理，得到所述第一基准图像；所述第二预设规则至少设置有对所述第一预定图像的截取或裁剪处理的位置及大小。所述第一预定图像可以为本地图片、通过摄像头拍摄而得的图片、从其它终端/服务器中接收到的图片或从网络下载的图片。例如，第一预定图像可以为前述的图片A；应使用需求，可以对图片A进行全部截取，也可以进行部分截取。例如，对于200*200像素的图片A来说，可以按照1/5的比例(去除掉40个像素点)从像素点(0，0)位置处开始进行截取，截取后的图片为160*160像素，包括像素点(0，0)～(160，160)；或者，按照1/5的比例从像素点(5，5)位置处开始进行截取，截取后的图片为160*160像素，包括像素点(5，5)～(165，165)，根据实际使用情况而定。
步骤302：获取至少一个待检测图像，按照第一预设规则对所述待检测图像进行至少一个指定方向的第N次截取或裁剪处理，得到所述至少一个待检测图像中的第N个子图像；
这里，所述待检测图像可以为本地图片、通过摄像头拍摄而得的图片、从终端/服务器中接收到的图片或从网络下载的图片。所述第一预设规则至少设置有对所述至少一个待检测图像进行第N次截取或裁剪处理的位置及大小；所述第N个子图像与所述第一基准图像具有相同的大小，N为正整数。待检测图像的大小可以与第一预定图像的大小相同或不同。所述指定方向为图片的横向方向和纵向方向，优选的可以先进行横向方向上的截取，当从横向方向上截取下的子图像与第一基准图像不存在相似可能时再进行纵向方向上的截取。第N个子图像的大小需要与第一基准图像的大小相同。例如，图片B为200*200个像素点，以横向方向进行截取、每次平移图片 的一个像素点为例，进行第1次截取时按照1/5的比例(去除掉200*1/5＝40个像素点)从像素点(0，0)位置处开始进行截取，截取后的图片即第1个子图像为160*160像素，包括有像素点(0，0)～(160，160)；进行第2次截取时按照1/5的比例，从像素点(1，0)位置处开始进行截取，截取后的图片即第2个子图像为160*160像素，包括有像素点(1，0)～(161，160)，以此类推。
步骤303：分别获取所述第一基准图像和所述第N个子图像中的第一属性值和/或第二属性值，所述第一属性值用于表征图像的颜色分布信息，所述第二属性值用于表征图像的指纹信息；
这里，所述第一属性值为通过CF算法计算出的颜色分布直方图，第二属性值通过phash算法计算出的图像的指纹信息；计算第一基准图像、第N个子图像的颜色分布直方图和/或图像指纹信息。具体的计算颜色分布直方图、图像指纹信息的过程请参见现有相关说明，此处不赘述。
步骤304：依据第一基准图像、所述第N个子图像的第一属性值，确定第一数据，所述第一数据为第一基准图像与所述第N个子图像在颜色上的相似程度；
这里，通过余弦相似算法计算出第一基准图像与第N个子图像关于颜色分布直方图的相似度值，得到第一数据，并将第一数据缓存起来。当然第一数据也可以通过其他算法而得。
步骤305：获取第二数据，所述第二数据表征为在所述第N个子图像之前的N-1个子图像中每个子图像的第一数据中的最大值；
这里，在对待检测图像进行的前N-1次截取得到前N-1个子图像后，当计算出这前N-1个子图像中每个子图像与第一基准图像的相似度值后，均需要缓存起来，以备后续方案使用。本步骤中读取之前已缓存的前N-1个子图像对应的第一数据，并计算出其中的最大值作为第二数据。
步骤306：依据所述第一数据、第二数据、第一基准图像的第二属性值及所述第N个子图像的第二属性值，确定所述第N个子图像的第一属性值和/或第二属性值是否与所述第一基准图像的对应属性值满足预定条件；
本步骤中，在所述第二数据及所述第N个子图像的第一数据中，确定所述第N个子图像的第一数据是否为最大值；
如果第N个子图像的第一数据不为最大值，对所述至少一个待检测图像进行第二指定方向上的截取或裁剪处理，所述第N个子图像由对所述至少一个待检测图像进行的第一指定方向上的截取或裁剪处理而得，也就是改变了对待测图像的截取方向。本实施例中，以先进行横向截取为例，第N个子图像的第一数据不为最大值，说明第N个子图像不存在有与第一基准图像相似的可能，改变对待检测图像的截取方向，从待检测图像的纵向方向上进行截取，截取位置与比例均可以与进行横向方向截取时的相同，获得相应的子图像，判断该子图像是否与第一基准图像相似，此过程与前述的内容相类似，不再描述。
如果第N个子图像的第一数据为最大值时，确定所述第N个子图像的第一属性值与第一基准图像的第一属性值满足预定条件，认为在横向方向上对待检测图像进行截取而得第N个子图像很可能相似于第一基准图像或所述待检测图像很可能相似于第一基准图像；接着计算第N个子图像与第一基准图像在第二属性值上的差值；当所述第N个子图像的第二属性值和第一基准图像的第二属性值之差未超出第一预定范围如第N个子图像和第一基准图像的图像指纹信息的汉明距离小于等于M，确定所述第N个子图像的第二属性值与所述第一基准图像的第二属性值满足预定条件。当然，如果所述第N个子图像的第二属性值和第一基准图像的第二属性值之差超出第一预定范围如第N个子图像和第一基准图像的图像指纹信息的汉明距离大于M时，确定所述第N个子图像的第二属性值与所述第一基准图像的 第二属性值不满足预定条件。其中，M为正整数，如M＝5，也可以取其它值，依据实际情况而设定。
步骤307：当所述第N个子图像的第一属性值和/或第二属性值与所述第一基准图像的对应属性值满足预定条件时，识别出所述第N个子图像隶属的所述待检测图像与所述第一基准图像相匹配。
例如，当第N个子图像的第一数据大于第二数据、和/或第N个子图像和第一基准图像的图像指纹信息的汉明距离小于等于M，认为第N个子图像的第一属性值和/或第二属性值与所述第一基准图像的对应属性值满足预定条件，此时可以肯定所述第N个子图像隶属的所述待检测图像与所述第一基准图像相匹配即为相似的图片。
进一步的，
在确定所述第N个子图像的第一属性值和/或第二属性值与所述第一基准图像的对应属性值不满足预定条件之后，所述方法还包括：
判断对所述至少一个待检测图像进行的所述指定方向上的第N次截取或裁剪处理是否为在所述指定方向上的最后一次截取或裁剪处理；
如果对所述至少一个待检测图像进行的所述指定方向上的第N次截取或裁剪处理不为在第一指定方向上的最后一次截取或裁剪处理，对所述至少一个待检测图像进行所述指定方向上的第N+1次截取或裁剪处理，得到第N+1个子图像；获取所述第N+1个子图像中的第一属性值和/或第二属性值；当所述第N+1个子图像的第一属性值和/或第二属性值与所述第一基准图像的对应属性值满足预定条件时，识别出所述第N+1个子图像隶属的所述待检测图像与所述第一基准图像相匹配。这里，以先进行的横向方向上的截取为例，对待检测图像进行在所述指定方向上如横向方向上的第N+1次截取，得到第N+1个子图像，并通过CF算法得到第N+1个子图像的颜色分布直方图和/或通过phash算法得到图像指纹，再根据颜色分布直方图 和图像指纹，确定第N+1个子图像或其所属的待检测图像是否相似于第一基准图像，具体过程请参见前述对第N个子图像的相关说明。
如果对所述至少一个待检测图像进行的所述指定方向上的第N次截取或裁剪处理为在第一指定方向上的最后一次截取或裁剪处理，对所述至少一个待检测图像进行第二指定方向上的截取或裁剪处理。这里，以对待检测图像先进行横向方向(第一指定方向)截取为例，在横向方向上按照一定的比例逐一截取完成后，也没有判断出存在有与第一基准图像相似的子图像，那么需要在另一个指定方向如纵向方向(第二指定方向)上按照预定的比例及截取位置逐次进行截取，并比较在纵向上截取出的子图像是否与第一基准图像相似，如果仍然不相似，改变第一基准图像，可以通过对第一基准图像的不断截取来改变。例如，第一次选取的第一基准图像为图片A的像素点(5，5)～(165，165)这部分，第二次选取的第一基准图像可以为图片A的像素点(10，5)～(170，165)这部分，第二次选取这部分图像可以通过第一次选取的第一基准图像向横向方向平移5个像素点而得到。其中，对在纵向上截取得到子图像的后续处理与前述的对在横向方向上截取得到的子图像的后续处理相类似，此处不再赘述。
值得说明的是，对于第N个子图像、第一基准图像的第一属性值，以及第N个子图像、第一基准图像的第二属性值同时满足相应预定条件的这种情况，先通过颜色直方图的相似程度这一要素确定第N个子图像是否存在有与第一基准图像相似的可能，如果通过颜色直方图的相似程度确定第N个子图像存在有与第一基准图像相似的可能时，再通过图像指纹信息这一要素确定第N个子图像是否存在有与第一基准图像相似的可能，如果通过前述的两个要素都存在有相似的可能，则认为第N个子图像或其隶属的待检测图像与第一基准图像为相似的图片，这种两种算法相结合的方式可保证不漏掉相似度高的图片也可以保证或提高识别正确率。
本发明实施例中，先获取第一基准图像，再通过对待检测图像进行某次截取或裁剪处理得到一个子图像，当第一基准图像与子图像在颜色分布信息方面和/或图像指纹信息方面存在有相似时，认为待检测图像是与第一基准图像相似的图像。优选的，当第一基准图像与子图像在颜色分布信息方面与图像指纹信息方面均存在有相似时，认为待检测图像是与第一基准图像相似的图像。其中，利用CF算法得到颜色部分直方图，利用phash算法得到图像指纹信息，两种算法相结合可保证不漏掉相似度高的图片也可以保证或提高识别正确率。
实施例三
图4为本发明实施例的装置实施例一的组成结构示意图；如图4所示，所述装置包括：第一获取单元401、第二获取单元402、第三获取单元403、及第一识别单元404；其中，
第一获取单元401，配置为获取第一基准图像；
在实际应用中，可以从其它图片中查找到与图片A本身或图片A的部分图像相似度高的图片，如在其它图片中查找包括有图片A中某个对象如牡丹花的图片。当想要从其它图片中查找与图片A本身相似度高的图片时，第一基准图像为图片A本身。当想要从其它图片中查找与图片A的部分图像如牡丹花图像相似度高的图片时，第一基准图像为图片A中包括有牡丹花的部分图像。
所述第一获取单元401先确定第一预定图像，然后按照第二预设规则对第一预定图像进行截取或裁剪处理，得到所述第一基准图像；所述第二预设规则至少设置有对所述第一预定图像的截取或裁剪处理的位置及大小。所述第一预定图像可以为本地图片、通过摄像头拍摄而得的图片、从其它终端/服务器中接收到的图片或从网络下载的图片。例如，第一预定图像可以为前述的图片A；应使用需求，可以对图片A进行全部截取，也可 以进行部分截取。例如，对于200*200像素的图片A来说，可以按照1/5的比例(去除掉40个像素点)从像素点(0，0)位置处开始进行截取，截取后的图片为160*160像素，包括像素点(0，0)～(160，160)；或者，按照1/5的比例从像素点(5，5)位置处开始进行截取，截取后的图片为160*160像素，包括像素点(5，5)～(165，165)。
第二获取单元402，配置为获取至少一个待检测图像，按照第一预设规则对所述待检测图像进行至少一个指定方向的第N次截取或裁剪处理，得到所述至少一个待检测图像中的第N个子图像，N为正整数；
这里，所述待检测图像可以为本地图片、通过摄像头拍摄而得的图片、从其它终端/服务器中接收到的图片或从网络下载的图片。所述第一预设规则至少设置有对所述至少一个待检测图像进行第N次截取或裁剪处理的位置及大小；所述第N个子图像与所述第一基准图像具有相同的大小，N为正整数。待检测图像的大小可以与第一预定图像的大小相同或不同；所述指定方向为图片的横向方向和纵向方向，优选的第二获取单元402可以先进行横向方向上的截取，通过从横向方向上截取下的子图像与第一基准图像不存在相似可能时再进行纵向方向上的截取。第N个子图像的大小需要与第一基准图像的大小相同。例如，图片B为200*200个像素点，以横向方向、每次平移图片的一个像素点为例，第二获取单元402进行第1次截取时按照1/5的比例(去除掉200*1/5＝40个像素点)从像素点(0，0)位置处开始进行截取，截取后的图片即第1个子图像为160*160像素，包括有像素点(0，0)～(160，160)；第二获取单元402进行第2次截取时按照1/5的比例，从像素点(1，0)位置处开始进行截取，截取后的图片即第2个子图像为160*160像素，包括有像素点(1，0)～(161，160)，以此类推。
第三获取单元403，配置为分别获取所述第一基准图像和所述第N个 子图像中的第一属性值和/或第二属性值，所述第一属性值用于表征图像的颜色分布信息，所述第二属性值用于表征图像的指纹信息；
这里，所述第一属性值为通过CF算法计算出的颜色分布直方图，第二属性值通过phash算法计算出的图像的指纹信息；第三获取单元403计算第一基准图像、第N个子图像的颜色分布直方图和/或图像指纹信息。具体的计算颜色分布直方图、图像指纹信息的过程请参见现有相关说明，此处不赘述。
第一识别单元404，配置为当所述第N个子图像的第一属性值和/或第二属性值与所述第一基准图像的对应属性值满足预定条件时，识别出所述第N个子图像隶属的所述待检测图像与所述第一基准图像相匹配。
这里，当第一识别单元404确认第N个子图像的颜色分布直方图与第一基准图像的颜色分布直方图的相似度较高、和/或第N个子图像的图像指纹与第一基准图像的图像指纹相似度较高，确认第N个子图像隶属的待检测图像为与第一基准图像相似的图像。
本发明实施例中，先获取第一基准图像，再通过对待检测图像进行某次截取或裁剪处理得到一个子图像，当第一基准图像与子图像在颜色分布信息方面和/或图像指纹信息方面存在有相似时，认为待检测图像是与第一基准图像相似的图像。其中，利用CF算法得到颜色部分直方图，利用phash算法得到图像指纹信息，可保证不漏掉相似度高的图片也可以保证或提高识别正确率。
实施例四
图5为本发明实施例的装置实施例二的组成结构示意图；如图5所示，所述装置包括：第一获取单元501、第二获取单元502、第三获取单元503、及第一识别单元504；其中，
第一获取单元501，配置为获取第一基准图像；
在实际应用中，可以从其它图片中查找到与图片A本身或图片A的部分图像相似度高的图片，如在其它图片中查找包括有图片A中某个对象如牡丹花的图片。当想要从其它图片中查找与图片A本身相似度高的图片时，第一基准图像为图片A本身。当想要从其它图片中查找与图片A的部分图像如牡丹花图像相似度高的图片时，第一基准图像为图片A中包括有牡丹花的部分图像。
所述第一获取单元501先确定第一预定图像，然后按照第二预设规则对第一预定图像进行截取或裁剪处理，得到所述第一基准图像；所述第二预设规则至少设置有对所述第一预定图像的截取或裁剪处理的位置及大小。所述第一预定图像可以为本地图片、通过摄像头拍摄而得的图片、从其它终端/服务器中接收到的图片或从网络下载的图片。例如，第一预定图像可以为前述的图片A；应使用需求，可以对图片A进行全部截取，也可以进行部分截取。例如，对于200*200像素的图片A来说，可以按照1/5的比例(去除掉40个像素点)从像素点(0，0)位置处开始进行截取，截取后的图片为160*160像素，包括像素点(0，0)～(160，160)；或者，按照1/5的比例从像素点(5，5)位置处开始进行截取，截取后的图片为160*160像素，包括像素点(5，5)～(165，165)。
第二获取单元502，配置为获取至少一个待检测图像，按照第一预设规则对所述待检测图像进行至少一个指定方向的第N次截取或裁剪处理，得到所述至少一个待检测图像中的第N个子图像，N为正整数；
这里，所述待检测图像可以为本地图片、通过摄像头拍摄而得的图片、从其它终端/服务器中接收到的图片或从网络下载的图片。所述第一预设规则至少设置有对所述至少一个待检测图像进行第N次截取或裁剪处理的位置及大小；所述第N个子图像与所述第一基准图像具有相同的大小，N为正整数。待检测图像的大小可以与第一预定图像的大小相同或不同；所述 指定方向为图片的横向方向和纵向方向。优选的，第二获取单元502可以先进行横向方向上的截取，通过从横向方向上截取下的子图像与第一基准图像不存在相似可能时再进行纵向方向上的截取。第N个子图像的大小需要与第一基准图像的大小相同。例如，图片B为200*200个像素点，以横向方向、每次平移图片的一个像素点为例，第二获取单元502进行第1次截取时按照1/5的比例(去除掉200*1/5＝40个像素点)从像素点(0，0)位置处开始进行截取，截取后的图片即第1个子图像为160*160像素，包括有像素点(0，0)～(160，160)；第二获取单元502进行第2次截取时按照1/5的比例，从像素点(1，0)位置处开始进行截取，截取后的图片即第2个子图像为160*160像素，包括有像素点(1，0)～(161，160)，以此类推。
第三获取单元503，配置为分别获取所述第一基准图像和所述第N个子图像中的第一属性值和/或第二属性值，所述第一属性值用于表征图像的颜色分布信息，所述第二属性值用于表征图像的指纹信息；
这里，所述第一属性值为通过CF算法计算出的颜色分布直方图，第二属性值通过phash算法计算出的图像的指纹信息；第三获取单元503计算第一基准图像、第N个子图像的颜色分布直方图和/或图像指纹信息。具体的计算颜色分布直方图、图像指纹信息的过程请参见现有相关说明，此处不赘述。
第三获取单元503，还配置为依据第一基准图像、所述第N个子图像的第一属性值，确定第一数据，所述第一数据为第一基准图像与所述第N个子图像在颜色上的相似程度；获取第二数据，所述第二数据为在所述第N个子图像之前的N-1个子图像中每个子图像的第一数据中的最大值；
这里，第三获取单元503依据第一基准图像、第N个子图像的颜色分布直方图，通过余弦相似算法计算出第一基准图像与第N个子图像关于颜 色分布直方图的相似度值，缓存到缓存单元(图5中未示意出)中。当然也可以通过其他算法而得。当第二获取单元502对待检测图像进行的前N-1次截取得到前N-1个子图像后，第三获取单元503计算出这前N-1个子图像中每个子图像与第一基准图像的相似度值，并缓存到缓存单元中，以备后续方案使用。所述第三获取单元503读取之前缓存在缓存单元中的前N-1个子图像对应的第一数据，并计算出其中的最大值作为第二数据。
所述第一识别单元504，配置为依据所述第一数据、第二数据、第一基准图像的第二属性值及所述第N个子图像的第二属性值，确定所述第N个子图像的第一属性值和/或第二属性值是否与所述第一基准图像的对应属性值满足预定条件。
进一步的，所述第一识别单元504在所述第二数据及所述第N个子图像的第一数据中，确定所述第N个子图像的第一数据是否为最大值；
如果第N个子图像的第一数据不为最大值，触发第二获取单元502对所述至少一个待检测图像进行第二指定方向上的截取或裁剪处理，所述第N个子图像由对所述至少一个待检测图像进行的第一指定方向上的截取或裁剪处理而得，也就是改变了对待测图像的截取方向。本实施例中，以先进行横向截取为例，第N个子图像的第一数据不为最大值，说明第N个子图像不存在有与第一基准图像相似的可能，改变对待检测图像的截取方向，从待检测图像的纵向方向上进行截取，并获得相应的子图像，并通过第一识别单元504判断该子图像是否与第一基准图像相似，此过程与前述的内容相类似，不再描述。
如果所述第一识别单元504确认第N个子图像的第一数据为最大值时，确定所述第N个子图像的第一属性值与第一基准图像的第一属性值满足预定条件，认为在横向方向上对待检测图像进行截取而得第N个子图像很可能相似于第一基准图像或所述待检测图像很可能相似于第一基准图像；接 着第一识别单元504计算第N个子图像与第一基准图像在第二属性值上的差值；当所述第N个子图像的第二属性值和第一基准图像的第二属性值之差未超出第一预定范围如第N个子图像和第一基准图像的图像指纹信息的汉明距离小于等于M，确定所述第N个子图像的第二属性值与所述第一基准图像的第二属性值满足预定条件。当然，如果第一识别单元504计算出所述第N个子图像的第二属性值和第一基准图像的第二属性值之差超出第一预定范围如第N个子图像和第一基准图像的图像指纹信息的汉明距离大于M时，确定所述第N个子图像的第二属性值与所述第一基准图像的第二属性值不满足预定条件。其中，M为正整数，如M＝5，也可以取其它值，依据实际情况而设定。
第一识别单元504，配置为当所述第N个子图像的第一属性值和/或第二属性值与所述第一基准图像的对应属性值满足预定条件时，识别出所述第N个子图像隶属的所述待检测图像与所述第一基准图像相匹配。
进一步的，
在第一识别单元504确定所述第N个子图像的第一属性值和/或第二属性值与所述第一基准图像的对应属性值不满足预定条件之后，
第一识别单元504判断对所述至少一个待检测图像进行的所述指定方向上的第N次截取或裁剪处理是否为在所述指定方向上的最后一次截取或裁剪处理；
如果第一识别单元504判断为对所述至少一个待检测图像进行的所述指定方向上的第N次截取或裁剪处理不为在第一指定方向上的最后一次截取或裁剪处理，触发第二获取单元502对所述至少一个待检测图像进行所述指定方向上的第N+1次截取或裁剪处理，得到第N+1个子图像；相应的，第三获取单元503获取所述第N+1个子图像中的第一属性值和/或第二属性值；第一识别单元504确认为所述第N+1个子图像的第一属性值和/或第二 属性值与所述第一基准图像的对应属性值满足预定条件时，识别出所述第N+1个子图像隶属的所述待检测图像与所述第一基准图像相匹配。这里，第二获取单元502对待检测图像进行在所述指定方向上如横向方向上的第N+1次截取，得到第N+1个子图像，第三获取单元503通过CF算法得到第N+1个子图像的颜色分布直方图和/或通过phash算法得到图像指纹，第一识别单元504再根据颜色分布直方图和图像指纹，确定第N+1个子图像或其所属的待检测图像是否相似于第一基准图像，具体过程请参见前述对第N个子图像的相关说明。
如果第一识别单元504判断为对所述至少一个待检测图像进行的所述指定方向上的第N次截取或裁剪处理为在第一指定方向上的最后一次截取或裁剪处理，触发第二获取单元502对所述至少一个待检测图像进行第二指定方向上的截取或裁剪处理。这里，以对待检测图像先进行第一指定方向如横向方向截取为例，当第二获取单元502在横向方向上按照一定的比例逐一截取完成后，第一识别单元504也没有判断出与第一基准图像相似的子图像，那么触发第二获取单元502在第二指定方向如纵向方向上按照预定的比例及截取位置逐次进行截取，第一识别单元504比较在纵向上截取出的子图像是否与第一基准图像相似，如果仍然不相似，触发第一切换单元(图5中未示意出)改变第一基准图像，可以通过对第一基准图像的不断截取来改变。例如，第一次选取的第一基准图像为图片A的像素点(5，5)～(165，165)这部分，第二次选取的第一基准图像可以为图片A的像素点(10，5)～(170，165)这部分，第二次选取这部分图像可以通过第一次选取的第一基准图像向横向方向平移5个像素点而得到。其中，对在纵向上截取得到子图像的后续处理与前述的对在横向方向上截取得到的子图像的后续处理相类似，此处不再赘述。
值得说明的是，对于第N个子图像的、第一基准图像的第一属性值， 以及第N个子图像、第一基准图像的第二属性值同时满足相应预定条件的这种情况，先通过颜色直方图的相似程度这一要素确定第N个子图像是否存在有与第一基准图像相似的可能，如果通过颜色直方图的相似程度确定第N个子图像存在有与第一基准图像相似的可能时，再通过图像指纹信息这一要素确定第N个子图像是否存在有与第一基准图像相似的可能，如果通过前述的两个要素都存在有相似的可能，则认为第N个子图像或其隶属的待检测图像与第一基准图像为相似的图片，这种两种算法相结合的方式可保证不漏掉相似度高的图片也可以保证或提高识别正确率。
本发明实施例中，先获取第一基准图像，再通过对待检测图像进行某次截取或裁剪处理得到一个子图像，当第一基准图像与子图像在颜色分布信息方面和/或图像指纹信息方面存在有相似时，认为待检测图像是与第一基准图像相似的图像。优选的，当第一基准图像与子图像在颜色分布信息方面与图像指纹信息方面均存在有相似时，认为待检测图像是与第一基准图像相似的图像。其中，利用CF算法得到颜色部分直方图，利用phash算法得到图像指纹信息，两种算法相结合可保证不漏掉相似度高的图片也可以保证或提高识别正确率。
下面结合图6(a)～(c)所示的应用场景对本发明实施例作进一步的说明。如图6(c)所示，
步骤700：获取第一基准图像；
如图6(a)所示，假定终端支持像素为200*200的图片A(第一预定图像)的全屏显示，图片A中包括有用户1，按照1/5的比例(去除掉40个像素点)从像素点(0，0)位置处(即图片的左上角)开始进行截取，截取后的图片即第一基准图像为160*160像素，包括像素点(0，0)～(160，160)(如图7(a)中的黑灰框所示)。
步骤701：获取至少一个待检测图像；
这里，待检测图像为如图6(b)所示的图片B。
步骤702：按照第一预设规则对所述待检测图像进行横向方向的第N次截取或裁剪处理，得到当前子图像(第N个子图像)；
这里，待检测图像(通过拍摄而得)为如图6(b)所示的图片B(包括有一只手)，假定图片B本身也为200*200像素、也按照1/5的比例进行横向方向截取，每次截取比上一次均沿着横向的正方向平移一个像素点，如第1次从像素点(0，0)处开始进行1/5比例的截取，第2次从(1，0)像素点处开始进行1/5比例的截取，第3次从(2，0)像素点处开始进行1/5比例截取。
假定当前为第4次截取，得到的第4个子图像包括图片B的像素点(4，0)～(164，160)，如图6(b)中的黑灰框所示。
通过以下步骤来判断当前子图像(第N＝4子图像)所隶属的图片B是否为与第一基准图像相似的图像。
步骤703：通过CF算法计算出的第一基准图像、当前子图像的颜色分布直方图；通过phash算法计算出第一基准图像、当前子图像的图像指纹信息；
步骤704：依据第一基准图像、当前子图像的颜色分布直方图，计算第一基准图像与当前子图像在颜色分布上的相似度值；读取第二数据；
这里，可通过余弦相似算法得出这两个图像在颜色分布上的相似程度即第一数据。因为假定N＝4，在计算出第1个子图像～第3个子图像中每个子图像的第一数据后，都缓存至缓存单元中，此时读取缓存单元中的这些值，并计算出其中的最大值，第二数据即为该最大值。
步骤705：判断第二数据与第一数据的大小；
判断为第一数据大于第二数据时，执行步骤706；
判断为第一数据小于第二数据时，执行步骤709；
步骤706：计算当前子图像与第一基准图像在图像指纹信息上的差异；
如果第N个子图像和第一基准图像的图像指纹信息的汉明距离小于等于M如M＝5，认为在图像指纹信息这一要素上这两个图像的差异较小，执行步骤707；
如果第N个子图像和第一基准图像的图像指纹信息的汉明距离大于M，认为在图像指纹信息这一要素上这两图像的差异较大，执行步骤708；
步骤707：当前子图像所隶属的图片B与第一基准图像为相似的图像，流程结束；
这里，也可以认为第N个子图像为与第一基准图像相似的图像。
步骤708：判断对所述至少一个待检测图像进行的横向方向上的第N次截取或裁剪处理是否为在所述横向方向上的最后一次截取或裁剪处理；
如果判断为是，执行步骤709；
如果判断为否，执行步骤710；
步骤709：对所述至少一个待检测图像进行纵向方向上的截取或裁剪处理，得到在纵向方向上截取到的子图像，作为当前子图像并返回步骤703；
步骤710：对所述至少一个待检测图像进行横向方向上的第N+1次截取或裁剪处理，得到第N+1个子图像，作为当前子图像并返回步骤703。
由此可见，在本发明实施例中，从第一预定图像上截取或裁剪得到第一基准图像，再从待检测图像上进行第N次截取或裁剪得到第N个子图像，如果第N个子图像与第一基准图像在颜色分布直方图上存在有很高的相似程度，再从图像指纹信息方面确认第N个子图像与第一基准图像确认是否有相似。如果在颜色分布上和图像指纹上均存在有相似，则认为第N子图像所隶属的图像即待检测图像与第一基准图像为相似的图像。其中，利用CF算法得到颜色分布直方图，利用phash算法得到图像指纹，这两个算法相结合且CF算法作为初筛算法(初步筛选出与第一基准图像可能有相似的 图片)不会漏掉相似度很高的图片，phash算法作为进一步确认的方法，可以保证或提高识别正确率。
这里需要指出的是，前述装置可以位于PC这种电子设备，还可以位于PAD，平板电脑，手提电脑这种便携电子设备中、还可以位于如手机这种智能移动终端，不限于这里的描述。为实现各单元功能而合并为一或各单元功能分体设置的图像信息识别处理装置(电子设备)至少包括用于存储数据的数据库和用于数据处理的处理器，或者包括设置于服务器内的存储介质或独立设置的存储介质。
其中，对于用于数据处理的处理器而言，在执行处理时，可以采用微处理器、中央处理器(CPU，Central Processing Unit)、数字信号处理器(DSP，Digital Singnal Processor)或可编程逻辑阵列(FPGA，Field－Programmable Gate Array)实现；对于存储介质来说，包含操作指令，该操作指令可以为计算机可执行代码，通过所述操作指令来实现上述本发明实施例图像信息识别处理方法流程中的各个步骤。
该装置作为硬件实体S11的一个示例如图7所示。所述装置包括处理器31、存储介质32以及至少一个外部通信接口33；所述处理器31、存储介质32以及外部通信接口33均通过总线34连接。
这里需要指出的是：以上涉及图像信息识别处理装置的描述，与上述方法描述是类似的，同方法的有益效果描述，不做赘述。对于本发明装置实施例中未披露的技术细节，请参照本发明上述实施例的描述。
相应的，本发明实施例还提供一种计算机存储介质，其中存储有计算机可执行指令，该计算机可执行指令配置执行上述图像信息识别处理方法。
本发明实施例所述集成的模块如果以软件功能模块的形式实现并作为独立的产品销售或使用时，也可以存储在一个计算机可读取存储介质中。基于这样的理解，本领域内的技术人员应明白，本申请的实施例可提供为 方法、系统、或计算机程序产品。因此，本申请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，本申请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上实施的计算机程序产品的形式，所述存储介质包括但不限于U盘、移动硬盘、只读存储器(ROM，Read-Only Memory)、磁盘存储器、CD-ROM、光学存储器等。
本申请是根据本申请实施例的方法、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。
尽管已描述了本申请的优选实施例，但本领域内的技术人员一旦得知了基本创造性概念，则可对这些实施例做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申请范围的所有变更和修改。
工业实用性
采用本发明实施例，从第N个子图像和第一基准图像在第一属性值和第二属性值等方面的相似性入手，也就是兼顾到图像的颜色分布情况和图像指纹信息这两个方面对图像相似识别的影响；采用本发明实施例可有效提高识别正确率，减少相似图片被漏掉的概率。

Claims (19)

  1. 一种图像信息识别处理方法，所述方法包括：
    获取第一基准图像；
    获取至少一个待检测图像，按照第一预设规则对所述待检测图像进行至少一个指定方向的第N次截取或裁剪处理，得到所述至少一个待检测图像中的第N个子图像，N为正整数；
    分别获取所述第一基准图像和所述第N个子图像中的第一属性值和/或第二属性值，所述第一属性值用于表征图像的颜色分布信息，所述第二属性值用于表征图像的指纹信息；
    当所述第N个子图像的第一属性值和/或第二属性值与所述第一基准图像的对应属性值满足预定条件时，识别出所述第N个子图像隶属的所述待检测图像与所述第一基准图像相匹配。
  2. 根据权利要求1所述的方法，其中，所述获取第一基准图像，包括：
    确定第一预定图像；
    按照第二预设规则对第一预定图像进行截取或裁剪处理，得到所述第一基准图像；
    所述第二预设规则至少设置有对所述第一预定图像的截取或裁剪处理的位置及大小。
  3. 根据权利要求1所述的方法，其中，所述第一预设规则至少设置有对所述至少一个待检测图像进行第N次截取或裁剪处理的位置及大小；所述第N个子图像与所述第一基准图像具有相同的大小。
  4. 根据权利要求1至3中任一项所述的方法，其中，所述当所述第N个子图像的第一属性值和/或第二属性值与所述第一基准图像的对应属性值满足预定条件时，识别出所述第N个子图像隶属的所述待检测图像与所述第一基准图像相匹配，包括：
    依据第一基准图像、所述第N个子图像的第一属性值，确定第一数据，所述第一数据为第一基准图像与所述第N个子图像在颜色上的相似程度；
    获取第二数据，所述第二数据为在所述第N个子图像之前的N-1个子图像中每个子图像的第一数据中的最大值；
    依据所述第一数据、第二数据、第一基准图像的第二属性值及所述第N个子图像的第二属性值，确定所述第N个子图像的第一属性值和/或第二属性值是否与所述第一基准图像的对应属性值满足预定条件。
  5. 根据权利要求4所述的方法，其中，所述依据所述第一数据、第二数据、第一基准图像的第二属性值及所述第N个子图像的第二属性值，确定所述第N个子图像的第一属性值和/或第二属性值是否与所述第一基准图像的对应属性值满足预定条件，包括：
    在所述第二数据及所述第N个子图像的第一数据中，确定所述第N个子图像的第一数据是否为最大值；
    确定第N个子图像的第一数据为最大值时，确定所述第N个子图像的第一属性值与第一基准图像的第一属性值满足预定条件；
    计算第N个子图像与第一基准图像在第二属性值上的差值；
    如果所述第N个子图像的第二属性值和第一基准图像的第二属性值之差未超出第一预定范围时，确定所述第N个子图像的第二属性值与所述第一基准图像的第二属性值满足预定条件。
  6. 根据权利要求5所述的方法，其中，所述方法还包括：
    如果所述第N个子图像的第二属性值和第一基准图像的第二属性值之差超出第一预定范围时，确定所述第N个子图像的第二属性值与所述第一基准图像的第二属性值不满足预定条件。
  7. 根据权利要求6所述的方法，其中，在确定所述第N个子图像的第一属性值和/或第二属性值与所述第一基准图像的对应属性值不满足预定条 件之后，所述方法还包括：
    判断对所述至少一个待检测图像进行的所述指定方向上的第N次截取或裁剪处理是否为在所述指定方向上的最后一次截取或裁剪处理；
    如果否，对所述至少一个待检测图像进行所述指定方向上的第N+1次截取或裁剪处理，得到第N+1个子图像；
    获取所述第N+1个子图像中的第一属性值和/或第二属性值；
    当所述第N+1个子图像的第一属性值和/或第二属性值与所述第一基准图像的对应属性值满足预定条件时，识别出所述第N+1个子图像隶属的所述待检测图像与所述第一基准图像相匹配。
  8. 根据权利要求7所述的方法，其中，所述方法还包括：
    如果判断为对所述至少一个待检测图像进行的所述指定方向上的第N次截取或裁剪处理为在第一指定方向上的最后一次截取或裁剪处理，对所述至少一个待检测图像进行第二指定方向上的截取或裁剪处理。
  9. 根据权利要求5所述的方法，其中，所述方法还包括：
    当确定所述第N个子图像的第一数据不为最大值时，
    对所述至少一个待检测图像进行第二指定方向上的截取或裁剪处理，所述第N个子图像由对所述至少一个待检测图像进行的第一指定方向上的截取或裁剪处理而得。
  10. 一种图像信息识别处理装置，所述装置包括：
    第一获取单元，配置为获取第一基准图像；
    第二获取单元，配置为获取至少一个待检测图像，按照第一预设规则对所述待检测图像进行至少一个指定方向的第N次截取或裁剪处理，得到所述至少一个待检测图像中的第N个子图像，N为正整数；
    第三获取单元，配置为分别获取所述第一基准图像和所述第N个子图像中的第一属性值和/或第二属性值，所述第一属性值用于表征图像的颜色 分布信息，所述第二属性值用于表征图像的指纹信息；
    第一识别单元，配置为当所述第N个子图像的第一属性值和/或第二属性值与所述第一基准图像的对应属性值满足预定条件时，识别出所述第N个子图像隶属的所述待检测图像与所述第一基准图像相匹配。
  11. 根据权利要求10所述的装置，其中，所述第一获取单元，配置为
    确定第一预定图像；
    按照第二预设规则对第一预定图像进行截取或裁剪处理，得到所述第一基准图像；
    所述第二预设规则至少设置有对所述第一预定图像的截取或裁剪处理的位置及大小。
  12. 根据权利要求10所述的装置，其中，所述第一预设规则至少设置有对所述至少一个待检测图像进行第N次截取或裁剪处理的位置及大小；所述第N个子图像与所述第一基准图像具有相同的大小。
  13. 根据权利要求10至12任一项所述的装置，其中，
    所述第三获取单元，配置为依据第一基准图像、所述第N个子图像的第一属性值，确定第一数据，所述第一数据为第一基准图像与所述第N个子图像在颜色上的相似程度；
    获取第二数据，所述第二数据为在所述第N个子图像之前的N-1个子图像中每个子图像的第一数据中的最大值；
    所述第一识别单元，配置为依据所述第一数据、第二数据、第一基准图像的第二属性值及所述第N个子图像的第二属性值，确定所述第N个子图像的第一属性值和/或第二属性值是否与所述第一基准图像的对应属性值满足预定条件。
  14. 根据权利要求13所述的装置，其中，所述第一识别单元，还配置为：
    在所述第二数据及所述第N个子图像的第一数据中，确定所述第N个子图像的第一数据是否为最大值；
    确定第N个子图像的第一数据为最大值时，确定所述第N个子图像的第一属性值与第一基准图像的第一属性值满足预定条件；
    计算第N个子图像与第一基准图像在第二属性值上的差值；
    如果所述第N个子图像的第二属性值和第一基准图像的第二属性值之差未超出第一预定范围时，确定所述第N个子图像的第二属性值与所述第一基准图像的第二属性值满足预定条件。
  15. 根据权利要求14所述的装置，其中，所述第一识别单元，还配置为：
    如果所述第N个子图像的第二属性值和第一基准图像的第二属性值之差超出第一预定范围时，确定所述第N个子图像的第二属性值与所述第一基准图像的第二属性值不满足预定条件。
  16. 根据权利要求15所述的装置，其中，所述第一识别单元，配置为：判断对所述至少一个待检测图像进行的所述指定方向上的第N次截取或裁剪处理是否为在所述指定方向上的最后一次截取或裁剪处理；
    如果否，触发第二获取单元；
    相应的，所述第二获取单元，配置为对所述至少一个待检测图像进行所述指定方向上的第N+1次截取或裁剪处理，得到第N+1个子图像；
    所述第三获取单元，配置为获取所述第N+1个子图像中的第一属性值和/或第二属性值；
    所述第一识别单元，配置为当所述第N+1个子图像的第一属性值和/或第二属性值与所述第一基准图像的对应属性值满足预定条件时，识别出所述第N+1个子图像隶属的所述待检测图像与所述第一基准图像相匹配。
  17. 根据权利要求16所述的装置，其中，所述第一识别单元，还配置 为：
    如果判断为对所述至少一个待检测图像进行的所述指定方向上的第N次截取或裁剪处理为在第一指定方向上的最后一次截取或裁剪处理，触发第二获取单元；
    相应的，所述第二获取单元，配置为对所述至少一个待检测图像进行第二指定方向上的截取或裁剪处理。
  18. 根据权利要求14所述的装置，其中，所述第一识别单元，还配置为：
    当确定所述第N个子图像的第一数据不为最大值时，触发第二获取单元；
    相应的，所述第二获取单元，配置为对所述至少一个待检测图像进行第二指定方向上的截取或裁剪处理，所述第N个子图像由对所述至少一个待检测图像进行的第一指定方向上的截取或裁剪处理而得。
  19. 一种计算机存储介质，其中存储有计算机可执行指令，该计算机可执行指令配置执行上述权利要求1至9任一项所述的图像信息识别处理方法。
PCT/CN2016/080564 2015-09-30 2016-04-28 一种图像信息识别处理方法及装置、计算机存储介质 WO2017054442A1 (zh)

Priority Applications (2)

Application Number Priority Date Filing Date Title
EP16850067.6A EP3273388A4 (en) 2015-09-30 2016-04-28 Image information recognition processing method and device, and computer storage medium
US15/729,935 US10438086B2 (en) 2015-09-30 2017-10-11 Image information recognition processing method and device, and computer storage medium

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201510642022.2 2015-09-30
CN201510642022.2A CN106560840B (zh) 2015-09-30 2015-09-30 一种图像信息识别处理方法及装置

Related Child Applications (1)

Application Number Title Priority Date Filing Date
US15/729,935 Continuation US10438086B2 (en) 2015-09-30 2017-10-11 Image information recognition processing method and device, and computer storage medium

Publications (1)

Publication Number Publication Date
WO2017054442A1 true WO2017054442A1 (zh) 2017-04-06

Family

ID=58422666

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2016/080564 WO2017054442A1 (zh) 2015-09-30 2016-04-28 一种图像信息识别处理方法及装置、计算机存储介质

Country Status (4)

Country Link
US (1) US10438086B2 (zh)
EP (1) EP3273388A4 (zh)
CN (1) CN106560840B (zh)
WO (1) WO2017054442A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111915550A (zh) * 2019-05-09 2020-11-10 杭州海康威视数字技术股份有限公司 图像质量检测方法、检测设备和存储介质

Families Citing this family (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106560840B (zh) * 2015-09-30 2019-08-13 腾讯科技（深圳）有限公司 一种图像信息识别处理方法及装置
CN107516105B (zh) * 2017-07-20 2020-06-16 阿里巴巴集团控股有限公司 图像处理方法及装置
CN108154031B (zh) * 2018-01-17 2021-08-06 腾讯科技（深圳）有限公司 伪装应用程序的识别方法、装置、存储介质和电子装置
WO2019205069A1 (en) * 2018-04-27 2019-10-31 Beijing Didi Infinity Technology And Development Co., Ltd. Systems and methods for updating 3d model of building
CN109344275B (zh) * 2018-08-21 2021-10-26 深圳市致远优学教育科技有限公司 基于图片识别的资源获取装置及方法
CN110287861B (zh) * 2019-06-21 2022-01-07 Oppo广东移动通信有限公司 指纹识别方法、装置、存储介质及电子设备
CN110633743A (zh) * 2019-09-09 2019-12-31 南昌市微轲联信息技术有限公司 一种基于邻域支撑的图像匹配方法
CN111405314B (zh) * 2020-03-09 2021-06-15 腾讯科技（深圳）有限公司 一种信息处理方法、装置、设备及存储介质
CN111753720A (zh) * 2020-06-24 2020-10-09 上海依图网络科技有限公司 一种指纹匹配方法及装置
CN112132146A (zh) * 2020-08-14 2020-12-25 北京三快在线科技有限公司 图像裁切模型的训练方法、装置及图像裁切方法、装置

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20020001036A1 (en) * 2000-03-14 2002-01-03 Naoto Kinjo Digital camera and image processing method
US20040252768A1 (en) * 2003-06-10 2004-12-16 Yoshinori Suzuki Computing apparatus and encoding program
CN103793697A (zh) * 2014-02-17 2014-05-14 北京旷视科技有限公司 一种人脸图像的身份标注方法及人脸身份识别方法
CN104217222A (zh) * 2014-09-25 2014-12-17 中国科学院自动化研究所 一种基于随机采样哈希表示的图像匹配方法
CN104463151A (zh) * 2015-01-05 2015-03-25 北京邮电大学 基于主颜色直方图和空间位置信息融合的快速目标匹配方法
CN104820996A (zh) * 2015-05-11 2015-08-05 河海大学常州校区 一种基于视频的自适应分块的目标跟踪方法
CN104881668A (zh) * 2015-05-13 2015-09-02 中国科学院计算技术研究所 一种基于代表性局部模式的图像指纹提取方法及系统

Family Cites Families (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7888621B2 (en) * 2006-09-29 2011-02-15 International Paper Co. Systems and methods for automatically adjusting the operational parameters of a laser cutter in a package processing environment
US7991189B2 (en) * 2007-03-28 2011-08-02 Sharp Kabushiki Kaisha Image processing apparatus, image forming apparatus, image processing system, and image processing method
US8260061B2 (en) * 2007-09-21 2012-09-04 Sharp Kabushiki Kaisha Image data output processing apparatus and image data output processing method
CN103105234B (zh) * 2012-01-12 2016-05-11 杭州美盛红外光电技术有限公司 热像装置和热像规范拍摄方法
US10685234B2 (en) * 2012-03-31 2020-06-16 Xerox Corporation Automatic and semi-automatic metadata generation via inheritance in homogeneous and heterogeneous environments
CN103617625B (zh) * 2013-12-13 2017-05-24 中国气象局北京城市气象研究所 一种图像匹配方法及图像匹配装置
CN103984776B (zh) * 2014-06-05 2017-05-03 北京奇虎科技有限公司 一种识别重复图片的方法、图片搜索去重方法及其装置
CN106560840B (zh) * 2015-09-30 2019-08-13 腾讯科技（深圳）有限公司 一种图像信息识别处理方法及装置

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20020001036A1 (en) * 2000-03-14 2002-01-03 Naoto Kinjo Digital camera and image processing method
US20040252768A1 (en) * 2003-06-10 2004-12-16 Yoshinori Suzuki Computing apparatus and encoding program
CN103793697A (zh) * 2014-02-17 2014-05-14 北京旷视科技有限公司 一种人脸图像的身份标注方法及人脸身份识别方法
CN104217222A (zh) * 2014-09-25 2014-12-17 中国科学院自动化研究所 一种基于随机采样哈希表示的图像匹配方法
CN104463151A (zh) * 2015-01-05 2015-03-25 北京邮电大学 基于主颜色直方图和空间位置信息融合的快速目标匹配方法
CN104820996A (zh) * 2015-05-11 2015-08-05 河海大学常州校区 一种基于视频的自适应分块的目标跟踪方法
CN104881668A (zh) * 2015-05-13 2015-09-02 中国科学院计算技术研究所 一种基于代表性局部模式的图像指纹提取方法及系统

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
See also references of EP3273388A4 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111915550A (zh) * 2019-05-09 2020-11-10 杭州海康威视数字技术股份有限公司 图像质量检测方法、检测设备和存储介质
CN111915550B (zh) * 2019-05-09 2024-03-29 杭州海康威视数字技术股份有限公司 图像质量检测方法、检测设备和存储介质

Also Published As

Publication number Publication date
US10438086B2 (en) 2019-10-08
CN106560840A (zh) 2017-04-12
EP3273388A1 (en) 2018-01-24
US20180225542A1 (en) 2018-08-09
CN106560840B (zh) 2019-08-13
EP3273388A4 (en) 2018-10-17

Similar Documents

Publication Publication Date Title
WO2017054442A1 (zh) 一种图像信息识别处理方法及装置、计算机存储介质
US11538232B2 (en) Tracker assisted image capture
EP2722850B1 (en) Method for generating thumbnail image and electronic device thereof
US20170032219A1 (en) Methods and devices for picture processing
WO2020093634A1 (zh) 基于人脸识别的照片添加方法、装置、终端及存储介质
US20150358549A1 (en) Image capturing parameter adjustment in preview mode
WO2018082308A1 (zh) 一种图像处理方法及终端
CN110427915B (zh) 用于输出信息的方法和装置
WO2021004186A1 (zh) 一种人脸采集方法、装置、系统、设备及介质
WO2015039575A1 (en) Method and system for performing image identification
WO2021227723A1 (zh) 目标检测方法、装置、计算机设备及可读存储介质
WO2018028253A1 (zh) 一种信息处理方法、装置及存储介质
US20230043154A1 (en) Restoring a video for improved watermark detection
CN111080665B (zh) 图像边框识别方法、装置、设备和计算机存储介质
WO2023005813A1 (zh) 图像方向调整方法、装置、存储介质及电子设备
CN107832598B (zh) 解锁控制方法及相关产品
CN110188782B (zh) 图像相似性确定方法、装置、电子设备及可读存储介质
CN116582653A (zh) 一种基于多摄像头数据融合的智能视频监控方法及系统
CN107770487B (zh) 一种特征提取并优选的方法、系统及终端设备
CN111210506A (zh) 一种三维还原方法、系统、终端设备和存储介质
US9955162B2 (en) Photo cluster detection and compression
CN111046831A (zh) 家禽识别方法、装置及服务器
CN108304840B (zh) 一种图像数据处理方法以及装置
US11647294B2 (en) Panoramic video data process
CN113158773B (zh) 一种活体检测模型的训练方法及训练装置

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: 16850067

Country of ref document: EP

Kind code of ref document: A1

REEP Request for entry into the european phase

Ref document number: 2016850067

Country of ref document: EP

NENP Non-entry into the national phase

Ref country code: DE