CN110443118A - Commodity recognition method, system and medium based on artificial feature - Google Patents
Commodity recognition method, system and medium based on artificial feature Download PDFInfo
- Publication number
- CN110443118A CN110443118A CN201910549890.4A CN201910549890A CN110443118A CN 110443118 A CN110443118 A CN 110443118A CN 201910549890 A CN201910549890 A CN 201910549890A CN 110443118 A CN110443118 A CN 110443118A
- Authority
- CN
- China
- Prior art keywords
- commodity
- feature
- result
- artificial feature
- artificial
- Prior art date
- Legal status (The legal status 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 status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06K—GRAPHICAL DATA READING; PRESENTATION OF DATA; RECORD CARRIERS; HANDLING RECORD CARRIERS
- G06K7/00—Methods or arrangements for sensing record carriers, e.g. for reading patterns
- G06K7/10—Methods or arrangements for sensing record carriers, e.g. for reading patterns by electromagnetic radiation, e.g. optical sensing; by corpuscular radiation
- G06K7/10544—Methods or arrangements for sensing record carriers, e.g. for reading patterns by electromagnetic radiation, e.g. optical sensing; by corpuscular radiation by scanning of the records by radiation in the optical part of the electromagnetic spectrum
- G06K7/10712—Fixed beam scanning
- G06K7/10722—Photodetector array or CCD scanning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- 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
-
- 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
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/35—Categorising the entire scene, e.g. birthday party or wedding scene
- G06V20/36—Indoor scenes
-
- G—PHYSICS
- G07—CHECKING-DEVICES
- G07G—REGISTERING THE RECEIPT OF CASH, VALUABLES, OR TOKENS
- G07G1/00—Cash registers
- G07G1/0036—Checkout procedures
- G07G1/0045—Checkout procedures with a code reader for reading of an identifying code of the article to be registered, e.g. barcode reader or radio-frequency identity [RFID] reader
-
- G—PHYSICS
- G07—CHECKING-DEVICES
- G07G—REGISTERING THE RECEIPT OF CASH, VALUABLES, OR TOKENS
- G07G1/00—Cash registers
- G07G1/12—Cash registers electronically operated
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- Biophysics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Electromagnetism (AREA)
- Toxicology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
Abstract
The present invention provides a kind of commodity recognition method based on artificial feature, system and media, include: image acquisition step: being shot to commodity are obtained in sales counter, Image Acquisition can all be triggered by opening and closing when cabinet door of selling goods, and the picture that Image Acquisition is obtained uploads；Commodity attribute step: according to the picture of upload, treating identification feature using the feature training pattern of acquisition and positioned, and obtains positioning result；Commodity identification clearance step: it is identified and is cleared according to the positioning result of acquisition and feature to be identified.The present invention, to the direct fixation and recognition of the apparent commodity of feature, solves the problems, such as algorithm excessively complexity and algorithm overlong time by neural network.
Description
Technical field
The present invention relates to image identification technical fields, and in particular, to commodity recognition method, system based on artificial feature
And medium.
Background technique
Paper " extracts [J] micro computer and application, 2015,34 (01): 50-52+ based on BP neural network two-dimension code area
58. " propose a kind of image in 2 D code fixation and recognition pretreatment system, can be quickly found out two-dimension code area.The system is main
It is divided into four parts: image procossing, feature extraction, sample training, test sample.Image procossing includes image enhancement and image opening and closing
Arithmetic operation.The working principle of the system is as follows: original image becomes binary image by image enhancement and opening and closing operations, and two
Being worth image will include one or more rectangular area, which passes through BP neural network, interference region is filtered, is obtained
Two-dimension code area.
The paper filters out interference region using BP neural network, rather than directly positions to two dimensional code, filtering
Accurately not can guarantee surely.In a practical situation, even if the system is accurately positioned to two dimensional code, the system is due to lacking post-processing
And correct decoding can not be done to two dimensional code.
CN109145816A (application number: 201810953349.5) discloses a kind of commodity recognition method and system, is related to
Commodity identify field.The commodity recognition method includes: that neural network module obtains product features and is transferred to the product features
Channel region pays attention to power module, wherein the product features include commodity correlated characteristic and commodity extraneous features；And channel region note
Meaning power module distinguishes the commodity correlated characteristic and the commodity extraneous features, and is at least transferred to the commodity correlated characteristic next
In a neural network module.
Summary of the invention
For the defects in the prior art, the object of the present invention is to provide a kind of commodity identification side based on artificial feature
Method, system and medium.
A kind of commodity recognition method based on artificial feature provided according to the present invention, comprising:
Image acquisition step: shooting to commodity are obtained in sales counter, and opening and closing when cabinet door of selling goods can all trigger
Image Acquisition, the picture that Image Acquisition is obtained upload；
Commodity attribute step: according to the picture of upload, identification feature is treated using the feature training pattern of acquisition and is determined
Position obtains positioning result；
Commodity identification clearance step: it is identified and is cleared according to the positioning result of acquisition and feature to be identified.
Preferably, further includes:
Feature preparation process: making artificial feature is attached on commodity, and commodity are put into sales counter；
Model training step: to the apparent commodity of feature, neural metwork training is carried out to product features, it is obvious to obtain feature
Commodity training pattern；For the commodity of artificial feature are added, neural metwork training is carried out to the artificial feature of such commodity, is obtained
Artificial feature commodity training pattern.
Preferably, the feature training pattern includes: the obvious commodity training pattern of feature, artificial feature commodity training mould
Type；
Feature to be identified includes: product features, artificial feature.
Preferably, the commodity identification clearance step, comprising:
Positioning result judgment step: judge whether positioning result is empty: if positioning result for sky, enters full figure check step
Suddenly it continues to execute, if positioning result is not sky, enters commodity identification step and continue to execute
Commodity identification step: according to the positioning result of acquisition, identifying product features and artificial feature, obtains commodity
Recognition result and artificial feature recognition result obtain commodity recognition result and enter complete if artificial feature recognition result is sky
Figure check step continues to execute, if artificial feature recognition result is not sky, obtains commodity recognition result and checks into commodity
Step continues to execute；
Full figure checks step: doing to full figure and turns distortion processing, be then decoded according to industrial decoding standard, decoding is tied
Fruit checks step into commodity as final commodity recognition result；
Commodity check step: according to the commodity recognition result and artificial feature recognition result of acquisition, will open the door and close the door
Merchandise news compares in the sales counter at two moment, obtains the final result of commodity identification.
Preferably, the feature preparation process:
The artificial feature includes: two dimensional code, point horizontal and vertical parity check code；
The sales counter includes: sales counter (1), camera (2) and headlamp (3)；
For camera (2), headlamp (3) and the commodity (4) all in sales counter (1), the camera is on sale
Above commodity (4) in counter (1), the artificial feature above feature and part commodity to cargo is shot.
Preferably, the commodity attribute step:
It according to the picture of upload, is positioned using artificial feature commodity training pattern, obtains all artificial spies in picture
The coordinate of sign is confined according to the coordinate pair artificial feature of all artificial features, and according to posting to the content in frame into
The interception of row picture, exports positioning result and screenshot；
For no commodity for pasting artificial feature, commodity are positioned using feature obvious commodity training pattern, it is defeated
Positioning result out.
If positioning result is sky, enters full figure check step and continue to execute, if positioning result is not sky, enter commodity
Identification step continues to execute.
Preferably, the commodity identification step:
The obvious commodity training pattern of feature is used according to the positioning result of acquisition for the commodity of no patch artificial feature
It is identified, obtains commodity recognition result；
For pasting the commodity of artificial feature, the screenshot of acquisition is decoded, comprising:
Operation and closed operation are opened to the artificial feature in screenshot, then decoded according to industry corresponding to the artificial feature
Standard is decoded, and obtains the first decoding result；
To artificial feature degree of the comparing enhancing in screenshot, then decodes and mark according to industry corresponding to the artificial feature
Standard is decoded, and obtains the second decoding result；
Artificial feature in screenshot is carried out to turn distortion correction, then decodes and marks according to industry corresponding to the artificial feature
Standard is decoded, and obtains the second decoding result；
According to the first, second of acquisition and third decoding result, if occurred in three kinds of decoding results of every screenshot
Two or more identical results, then the decoding result thinks artificial feature recognition result；If only there is a kind of decoding result, separately
Outer two methods can not decode, then taking the decoding result is artificial feature recognition result；
Judge whether artificial feature recognition result is empty: if artificial feature recognition result for sky, enters full figure check step
Suddenly it continues to execute；If artificial feature recognition result is not sky, obtains commodity recognition result and check step continuation into commodity
It executes；
It is described to judge whether artificial feature recognition result is empty:
If there is following situations, determine artificial feature recognition result for sky: three kinds of methods can not all decode；A kind of solution
Code result is sky, in addition two kinds of decoding result differences；Three kinds of decoding results are different；
It is described to open operation and closed operation:
The opening and closing operations are the methods of Morphological scale-space, first corrode the operation expanded afterwards and are referred to as out operation；
The closed operation refers to the operation for first expanding post-etching.
A kind of product identification system based on artificial feature provided according to the present invention, comprising:
Image capture module: shooting to commodity are obtained in sales counter, and opening and closing when cabinet door of selling goods can all trigger
Image Acquisition, the picture that Image Acquisition is obtained upload；
Commodity attribute module: according to the picture of upload, identification feature is treated using the feature training pattern of acquisition and is determined
Position obtains positioning result；
Commodity identification clearance module: it is identified and is cleared according to the positioning result of acquisition and feature to be identified.
Preferably, further includes:
Feature preparation module: making artificial feature is attached on commodity, and commodity are put into sales counter；
Model training module: to the apparent commodity of feature, neural metwork training is carried out to product features, it is obvious to obtain feature
Commodity training pattern；For the commodity of artificial feature are added, neural metwork training is carried out to the artificial feature of such commodity, is obtained
Artificial feature commodity training pattern；
The feature training pattern includes: the obvious commodity training pattern of feature, artificial feature commodity training pattern；
Feature to be identified includes: product features, artificial feature；
The commodity identification clearance module, comprising:
Positioning result judgment module: judge whether positioning result is empty: if positioning result for sky, enters full figure and checks mould
Block continues to execute, if positioning result is not sky, enters commodity identification module and continues to execute
Commodity identification module: according to the positioning result of acquisition, identifying product features and artificial feature, obtains commodity
Recognition result and artificial feature recognition result obtain commodity recognition result and call complete if artificial feature recognition result is sky
Figure check module obtains commodity recognition result and commodity is called to check module if artificial feature recognition result is not sky；
Full figure checks module: doing to full figure and turns distortion processing, be then decoded according to industrial decoding standard, decoding is tied
Fruit calls commodity to check module as final commodity recognition result；
Commodity check module: according to the commodity recognition result and artificial feature recognition result of acquisition, will open the door and close the door
Merchandise news compares in the sales counter at two moment, obtains the final result of commodity identification；
The feature preparation module:
The artificial feature includes: two dimensional code, point horizontal and vertical parity check code；
The sales counter includes: sales counter (1), camera (2) and headlamp (3)；
For camera (2), headlamp (3) and the commodity (4) all in sales counter (1), the camera is on sale
Above commodity (4) in counter (1), the artificial feature above feature and part commodity to cargo is shot；
The commodity attribute module:
It according to the picture of upload, is positioned using artificial feature commodity training pattern, obtains all artificial spies in picture
The coordinate of sign is confined according to the coordinate pair artificial feature of all artificial features, and according to posting to the content in frame into
The interception of row picture, exports positioning result and screenshot；
For no commodity for pasting artificial feature, commodity are positioned using feature obvious commodity training pattern, it is defeated
Positioning result out.
If positioning result is sky, full figure is called to check module, if positioning result is not sky, commodity is called to identify mould
Block；
The commodity identification module:
The obvious commodity training pattern of feature is used according to the positioning result of acquisition for the commodity of no patch artificial feature
It is identified, obtains commodity recognition result；
For pasting the commodity of artificial feature, the screenshot of acquisition is decoded, comprising:
Operation and closed operation are opened to the artificial feature in screenshot, then decoded according to industry corresponding to the artificial feature
Standard is decoded, and obtains the first decoding result；
To artificial feature degree of the comparing enhancing in screenshot, then decodes and mark according to industry corresponding to the artificial feature
Standard is decoded, and obtains the second decoding result；
Artificial feature in screenshot is carried out to turn distortion correction, then decodes and marks according to industry corresponding to the artificial feature
Standard is decoded, and obtains the second decoding result；
According to the first, second of acquisition and third decoding result, if occurred in three kinds of decoding results of every screenshot
Two or more identical results, then the decoding result thinks artificial feature recognition result；If only there is a kind of decoding result, separately
Outer two methods can not decode, then taking the decoding result is artificial feature recognition result；
Judge whether artificial feature recognition result is empty: if artificial feature recognition result calls full figure to check mould for sky
Block；If artificial feature recognition result is not sky, obtains commodity recognition result and commodity is called to check module；
It is described to judge whether artificial feature recognition result is empty:
If there is following situations, determine artificial feature recognition result for sky: three kinds of methods can not all decode；A kind of solution
Code result is sky, in addition two kinds of decoding result differences；Three kinds of decoding results are different；
It is described to open operation and closed operation:
The opening and closing operations are the methods of Morphological scale-space, first corrode the operation expanded afterwards and are referred to as out operation；
The closed operation refers to the operation for first expanding post-etching.
A kind of computer readable storage medium for being stored with computer program provided according to the present invention, which is characterized in that
Described in any item commodity recognition methods based on artificial feature among the above are realized when the computer program is executed by processor
The step of.
Compared with prior art, the present invention have it is following the utility model has the advantages that
1, the present invention by neural network to the direct fixation and recognition of the apparent commodity of feature, solve algorithm it is excessively complicated and
The problem of algorithm overlong time.
2, the present invention is identified by adding artificial feature to the unconspicuous commodity of feature by this medium of artificial feature
Commodity solve the problems, such as that part commodity can not fixation and recognition.
3, the present invention is by using a variety of methods such as machine vision and image procossing, compared to according to industrial standard directly into
Row identification solves the problems, such as that decoding rate is low and decoding accuracy rate is low.
Detailed description of the invention
Upon reading the detailed description of non-limiting embodiments with reference to the following drawings, other feature of the invention,
Objects and advantages will become more apparent upon:
Fig. 1 is the structural schematic diagram of sales counter provided by the invention.
Fig. 2 is the workflow block diagram representation of system provided by the invention.
Fig. 3 is algorithm flow schematic diagram provided by the invention.
Specific embodiment
The present invention is described in detail combined with specific embodiments below.Following embodiment will be helpful to the technology of this field
Personnel further understand the present invention, but the invention is not limited in any way.It should be pointed out that the ordinary skill of this field
For personnel, without departing from the inventive concept of the premise, several changes and improvements can also be made.These belong to the present invention
Protection scope.
A kind of commodity recognition method based on artificial feature provided according to the present invention, comprising:
Image acquisition step: shooting to commodity are obtained in sales counter, and opening and closing when cabinet door of selling goods can all trigger
Image Acquisition, the picture that Image Acquisition is obtained upload；
Commodity attribute step: according to the picture of upload, identification feature is treated using the feature training pattern of acquisition and is determined
Position obtains positioning result；
Commodity identification clearance step: it is identified and is cleared according to the positioning result of acquisition and feature to be identified.
Specifically, further includes:
Feature preparation process: making artificial feature is attached on commodity, and commodity are put into sales counter；
Model training step: to the apparent commodity of feature, neural metwork training is carried out to product features, it is obvious to obtain feature
Commodity training pattern；For the commodity of artificial feature are added, neural metwork training is carried out to the artificial feature of such commodity, is obtained
Artificial feature commodity training pattern.
Specifically, the feature training pattern includes: the obvious commodity training pattern of feature, artificial feature commodity training mould
Type；
Feature to be identified includes: product features, artificial feature.
Specifically, the commodity identification clearance step, comprising:
Positioning result judgment step: judge whether positioning result is empty: if positioning result for sky, enters full figure check step
Suddenly it continues to execute, if positioning result is not sky, enters commodity identification step and continue to execute
Commodity identification step: according to the positioning result of acquisition, identifying product features and artificial feature, obtains commodity
Recognition result and artificial feature recognition result obtain commodity recognition result and enter complete if artificial feature recognition result is sky
Figure check step continues to execute, if artificial feature recognition result is not sky, obtains commodity recognition result and checks into commodity
Step continues to execute；
Full figure checks step: doing to full figure and turns distortion processing, be then decoded according to industrial decoding standard, decoding is tied
Fruit checks step into commodity as final commodity recognition result；
Commodity check step: according to the commodity recognition result and artificial feature recognition result of acquisition, will open the door and close the door
Merchandise news compares in the sales counter at two moment, obtains the final result of commodity identification.
Specifically, the feature preparation process:
The artificial feature includes: two dimensional code, point horizontal and vertical parity check code；
The sales counter includes: sales counter (1), camera (2) and headlamp (3)；
For camera (2), headlamp (3) and the commodity (4) all in sales counter (1), the camera is on sale
Above commodity (4) in counter (1), the artificial feature above feature and part commodity to cargo is shot.
Specifically, the commodity attribute step:
It according to the picture of upload, is positioned using artificial feature commodity training pattern, obtains all artificial spies in picture
The coordinate of sign is confined according to the coordinate pair artificial feature of all artificial features, and according to posting to the content in frame into
The interception of row picture, exports positioning result and screenshot；
For no commodity for pasting artificial feature, commodity are positioned using feature obvious commodity training pattern, it is defeated
Positioning result out.
If positioning result is sky, enters full figure check step and continue to execute, if positioning result is not sky, enter commodity
Identification step continues to execute.
Specifically, the commodity identification step:
The obvious commodity training pattern of feature is used according to the positioning result of acquisition for the commodity of no patch artificial feature
It is identified, obtains commodity recognition result；
For pasting the commodity of artificial feature, the screenshot of acquisition is decoded, comprising:
Operation and closed operation are opened to the artificial feature in screenshot, then decoded according to industry corresponding to the artificial feature
Standard is decoded, and obtains the first decoding result；
To artificial feature degree of the comparing enhancing in screenshot, then decodes and mark according to industry corresponding to the artificial feature
Standard is decoded, and obtains the second decoding result；
Artificial feature in screenshot is carried out to turn distortion correction, then decodes and marks according to industry corresponding to the artificial feature
Standard is decoded, and obtains the second decoding result；
According to the first, second of acquisition and third decoding result, if occurred in three kinds of decoding results of every screenshot
Two or more identical results, then the decoding result thinks artificial feature recognition result；If only there is a kind of decoding result, separately
Outer two methods can not decode, then taking the decoding result is artificial feature recognition result；
Judge whether artificial feature recognition result is empty: if artificial feature recognition result for sky, enters full figure check step
Suddenly it continues to execute；If artificial feature recognition result is not sky, obtains commodity recognition result and check step continuation into commodity
It executes；
It is described to judge whether artificial feature recognition result is empty:
If there is following situations, determine artificial feature recognition result for sky: three kinds of methods can not all decode；A kind of solution
Code result is sky, in addition two kinds of decoding result differences；Three kinds of decoding results are different；
It is described to open operation and closed operation:
The opening and closing operations are the methods of Morphological scale-space, first corrode the operation expanded afterwards and are referred to as out operation；
The closed operation refers to the operation for first expanding post-etching.
A kind of product identification system based on artificial feature provided according to the present invention, comprising:
Image capture module: shooting to commodity are obtained in sales counter, and opening and closing when cabinet door of selling goods can all trigger
Image Acquisition, the picture that Image Acquisition is obtained upload；
Commodity attribute module: according to the picture of upload, identification feature is treated using the feature training pattern of acquisition and is determined
Position obtains positioning result；
Commodity identification clearance module: it is identified and is cleared according to the positioning result of acquisition and feature to be identified.
Specifically, further includes:
Feature preparation module: making artificial feature is attached on commodity, and commodity are put into sales counter；
Model training module: to the apparent commodity of feature, neural metwork training is carried out to product features, it is obvious to obtain feature
Commodity training pattern；For the commodity of artificial feature are added, neural metwork training is carried out to the artificial feature of such commodity, is obtained
Artificial feature commodity training pattern；
The feature training pattern includes: the obvious commodity training pattern of feature, artificial feature commodity training pattern；
Feature to be identified includes: product features, artificial feature；
The commodity identification clearance module, comprising:
Positioning result judgment module: judge whether positioning result is empty: if positioning result for sky, enters full figure and checks mould
Block continues to execute, if positioning result is not sky, enters commodity identification module and continues to execute
Commodity identification module: according to the positioning result of acquisition, identifying product features and artificial feature, obtains commodity
Recognition result and artificial feature recognition result obtain commodity recognition result and call complete if artificial feature recognition result is sky
Figure check module obtains commodity recognition result and commodity is called to check module if artificial feature recognition result is not sky；
Full figure checks module: doing to full figure and turns distortion processing, be then decoded according to industrial decoding standard, decoding is tied
Fruit calls commodity to check module as final commodity recognition result；
Commodity check module: according to the commodity recognition result and artificial feature recognition result of acquisition, will open the door and close the door
Merchandise news compares in the sales counter at two moment, obtains the final result of commodity identification；
The feature preparation module:
The artificial feature includes: two dimensional code, point horizontal and vertical parity check code；
The sales counter includes: sales counter (1), camera (2) and headlamp (3)；
For camera (2), headlamp (3) and the commodity (4) all in sales counter (1), the camera is on sale
Above commodity (4) in counter (1), the artificial feature above feature and part commodity to cargo is shot；
The commodity attribute module:
It according to the picture of upload, is positioned using artificial feature commodity training pattern, obtains all artificial spies in picture
The coordinate of sign is confined according to the coordinate pair artificial feature of all artificial features, and according to posting to the content in frame into
The interception of row picture, exports positioning result and screenshot；
For no commodity for pasting artificial feature, commodity are positioned using feature obvious commodity training pattern, it is defeated
Positioning result out.
If positioning result is sky, full figure is called to check module, if positioning result is not sky, commodity is called to identify mould
Block；
The commodity identification module:
The obvious commodity training pattern of feature is used according to the positioning result of acquisition for the commodity of no patch artificial feature
It is identified, obtains commodity recognition result；
For pasting the commodity of artificial feature, the screenshot of acquisition is decoded, comprising:
Operation and closed operation are opened to the artificial feature in screenshot, then decoded according to industry corresponding to the artificial feature
Standard is decoded, and obtains the first decoding result；
To artificial feature degree of the comparing enhancing in screenshot, then decodes and mark according to industry corresponding to the artificial feature
Standard is decoded, and obtains the second decoding result；
Artificial feature in screenshot is carried out to turn distortion correction, then decodes and marks according to industry corresponding to the artificial feature
Standard is decoded, and obtains the second decoding result；
According to the first, second of acquisition and third decoding result, if occurred in three kinds of decoding results of every screenshot
Two or more identical results, then the decoding result thinks artificial feature recognition result；If only there is a kind of decoding result, separately
Outer two methods can not decode, then taking the decoding result is artificial feature recognition result；
Judge whether artificial feature recognition result is empty: if artificial feature recognition result calls full figure to check mould for sky
Block；If artificial feature recognition result is not sky, obtains commodity recognition result and commodity is called to check module；
It is described to judge whether artificial feature recognition result is empty:
If there is following situations, determine artificial feature recognition result for sky: three kinds of methods can not all decode；A kind of solution
Code result is sky, in addition two kinds of decoding result differences；Three kinds of decoding results are different；
It is described to open operation and closed operation:
The opening and closing operations are the methods of Morphological scale-space, first corrode the operation expanded afterwards and are referred to as out operation；
The closed operation refers to the operation for first expanding post-etching.
A kind of computer readable storage medium for being stored with computer program provided according to the present invention, which is characterized in that
Described in any item commodity recognition methods based on artificial feature among the above are realized when the computer program is executed by processor
The step of.
Below by preference, the present invention is more specifically illustrated.
Preference 1:
As shown in Fig. 2, a kind of commodity recognition method based on artificial feature, comprising:
Step 1: the step 1 comprises the steps of:
Step 1.1: the preparation stage, making artificial feature (such as micro code two dimensional code) is attached on commodity, by commodity
It is put into the sales counter with specific structure, structure includes sales counter 1, camera 2, headlamp 3 and quotient as shown in Figure 1:
Product 4.Wherein, the camera 2, headlamp 3 and commodity 4 are all in sales counter 1.The camera above commodity 4,
The artificial feature above feature and part commodity to cargo is shot.The focal length of camera 2, wide-angle and apart from cargo
Distance parameter should be set in conjunction with concrete condition, guarantee to clearly photograph the feature above commodity.
Step 1.2: neural metwork training being carried out to the apparent commodity of feature, obtains its training pattern, such commodity will not
Artificial feature can be added.
Step 1.3: the commodity for artificial feature is added will carry out neural network instruction to the artificial feature of such commodity
Practice, obtains its training pattern.
Step 2: Image Acquisition.Camera 2 shoots commodity 4, and opening and closing when cabinet door of selling goods can all trigger
Picture is uploaded to cloud identification center and identified by Image Acquisition.
Step 3: positioning stage.The step 3 comprises the steps of:
Step 3.1: cloud identification center obtains the picture that will be identified.
Step 3.2: being positioned, obtained all artificial in picture using the training pattern of the artificial feature in step 1.3
The coordinate of feature, i.e. preloading training pattern, are loaded into neural network for picture.Network is general by processing acquisition picture prediction
Rate is characterized the region of code, exports the coordinate value in the region.For no commodity for pasting artificial feature, using in step 1.2
Training pattern directly commodity are positioned.
Step 3.3: being confined according to coordinate pair artificial feature, and picture is carried out to the content in frame according to posting and is cut
It takes.
Step 3.4: if the neural network positioning result in step 3.1 and 3.2 is sky, being directly entered step 5.
Step 4: cognitive phase.The step 4 comprises the steps of:
Step 4.1: for the commodity of no patch artificial feature, directly using the model of step 1.2 and according in step 3.2
Position identified, and obtain recognition result.
Step 4.2: the commodity for pasting artificial feature do following operation to the screenshot in step 3.3: (1) to screenshot
In artificial feature open operation and closed operation, be then decoded according to industrial decoding standard corresponding to the artificial feature.
(2) in screenshot artificial feature degree of comparing enhance, then according to industrial decoding standard corresponding to the artificial feature into
Row decoding.(3) artificial feature in screenshot is carried out turning distortion correction, is then decoded according to industry corresponding to the artificial feature
Standard is decoded.Opening and closing operations are the methods of Morphological scale-space, first corrode the operation expanded afterwards and are referred to as out operation.It has
Small objects are eliminated, in the effect of very thin place's separating objects and smooth larger object boundary.The operation for first expanding post-etching is referred to as
For closed operation.It has minuscule hole in filler body, connects the effect of adjacent object and smooth boundary.
Step 4.3: three kinds of decoding results in comparison step 4.2, if occurring two in three kinds of results of every screenshot
Or more identical result, then the result thinks the decoding result of artificial feature.If only there is one kind as a result, other two methods
It can not decode, then taking the result is the decoding result of artificial feature.If there is following situations, output is sky, and is entered step
5, otherwise entering step 6:(1) three kinds of methods can not all decode.(2) a kind of decoding result is sky, in addition two kinds of decoding results
It is different.(3) three kinds of decoding results are different.
Step 5: full figure check.Full figure is done and turns distortion processing, is then directly decoded according to industrial decoding standard, this
As a result as final result after step 4.3 or step 3.4.The industry decoding standard, such as artificial feature can be qr
Code two dimensional code, the two dimensional code have decoding standard, can also paste micro qr code two dimensional code, also there is corresponding decoding
Standard.Which type of artificial feature is pasted, just with which type of decoding standard.
Step 6: checking commodity.The recognition result for summarizing all commodity returns it to business backstage, then carries out cloud
Processing.Merchandise news in the cabinet at two moment of opening the door and close the door is compared, the final result of commodity identification is obtained.
As shown in figure 3, being the algorithm flow schematic diagram of step 3- step 6.
In the description of the present application, it is to be understood that term " on ", "front", "rear", "left", "right", " is erected at "lower"
Directly ", the orientation or positional relationship of the instructions such as "horizontal", "top", "bottom", "inner", "outside" is orientation based on the figure or position
Relationship is set, description the application is merely for convenience of and simplifies description, rather than the device or element of indication or suggestion meaning are necessary
It with specific orientation, is constructed and operated in a specific orientation, therefore should not be understood as the limitation to the application.
One skilled in the art will appreciate that in addition to realizing system provided by the invention in a manner of pure computer readable program code
It, completely can be by the way that method and step be carried out programming in logic come so that provided by the invention other than system, device and its modules
System, device and its modules are declined with logic gate, switch, specific integrated circuit, programmable logic controller (PLC) and insertion
The form of controller etc. realizes identical program.So system provided by the invention, device and its modules may be considered that
It is a kind of hardware component, and the knot that the module for realizing various programs for including in it can also be considered as in hardware component
Structure；It can also will be considered as realizing the module of various functions either the software program of implementation method can be Hardware Subdivision again
Structure in part.
Specific embodiments of the present invention are described above.It is to be appreciated that the invention is not limited to above-mentioned
Particular implementation, those skilled in the art can make a variety of changes or modify within the scope of the claims, this not shadow
Ring substantive content of the invention.In the absence of conflict, the feature in embodiments herein and embodiment can any phase
Mutually combination.
Claims (10)
1. a kind of commodity recognition method based on artificial feature characterized by comprising
Image acquisition step: shooting to commodity are obtained in sales counter, and image can all be triggered by opening and closing when cabinet door of selling goods
Acquisition, the picture that Image Acquisition is obtained upload；
Commodity attribute step: according to the picture of upload, identification feature is treated using the feature training pattern of acquisition and is positioned, is obtained
Obtain positioning result；
Commodity identification clearance step: it is identified and is cleared according to the positioning result of acquisition and feature to be identified.
2. the commodity recognition method according to claim 1 based on artificial feature, which is characterized in that further include:
Feature preparation process: making artificial feature is attached on commodity, and commodity are put into sales counter；
Model training step: to the apparent commodity of feature, neural metwork training is carried out to product features, obtains the obvious commodity of feature
Training pattern；For the commodity of artificial feature are added, neural metwork training is carried out to the artificial feature of such commodity, is obtained artificial
Feature commodity training pattern.
3. the commodity recognition method according to claim 2 based on artificial feature, which is characterized in that the feature training mould
Type includes: the obvious commodity training pattern of feature, artificial feature commodity training pattern；
Feature to be identified includes: product features, artificial feature.
4. the commodity recognition method according to claim 3 based on artificial feature, which is characterized in that the commodity identification is clear
Calculate step, comprising:
Positioning result judgment step: judge whether positioning result is empty: if positioning result for sky, enter full figure check step after
It is continuous to execute, if positioning result is not sky, enters commodity identification step and continue to execute
Commodity identification step: according to the positioning result of acquisition, identifying product features and artificial feature, obtains commodity identification
As a result it obtains commodity recognition result if artificial feature recognition result is sky with artificial feature recognition result and enters full figure and answer
It looks into step to continue to execute, if artificial feature recognition result is not sky, obtains commodity recognition result and check step into commodity
It continues to execute；
Full figure checks step: doing to full figure and turns distortion processing, is then decoded according to industrial decoding standard, decoding result is made
For final commodity recognition result, step is checked into commodity；
Commodity check step: according to the commodity recognition result and artificial feature recognition result of acquisition, will open the door and close the door two
Merchandise news compares in the sales counter at moment, obtains the final result of commodity identification.
5. the commodity recognition method according to claim 4 based on artificial feature, which is characterized in that the feature prepares step
It is rapid:
The artificial feature includes: two dimensional code, point horizontal and vertical parity check code；
The sales counter includes: sales counter (1), camera (2) and headlamp (3)；
Camera (2), headlamp (3) and the commodity (4) are all in sales counter (1), and the camera is in sales counter
(1) above the commodity (4) in, the artificial feature above feature and part commodity to cargo is shot.
6. the commodity recognition method according to claim 5 based on artificial feature, which is characterized in that the commodity attribute step
It is rapid:
It according to the picture of upload, is positioned using artificial feature commodity training pattern, obtains all artificial features in picture
Coordinate is confined according to the coordinate pair artificial feature of all artificial features, and carries out figure to the content in frame according to posting
Piece interception, exports positioning result and screenshot；
For no commodity for pasting artificial feature, commodity are positioned using feature obvious commodity training pattern, output is fixed
Position result.
If positioning result is sky, enters full figure check step and continue to execute, if positioning result is not sky, enters commodity and identify
Step continues to execute.
7. the commodity recognition method according to claim 6 based on artificial feature, which is characterized in that the commodity identification step
It is rapid:
The commodity of no patch artificial feature are carried out according to the positioning result of acquisition using the obvious commodity training pattern of feature
Identification obtains commodity recognition result；
For pasting the commodity of artificial feature, the screenshot of acquisition is decoded, comprising:
Operation and closed operation are opened to the artificial feature in screenshot, then according to industrial decoding standard corresponding to the artificial feature
It is decoded, obtains the first decoding result；
To in screenshot artificial feature degree of comparing enhancing, then according to industrial decoding standard corresponding to the artificial feature into
Row decoding, obtains the second decoding result；
Artificial feature in screenshot is carried out to turn distortion correction, then according to industrial decoding standard corresponding to the artificial feature into
Row decoding, obtains the second decoding result；
According to the first, second of acquisition and third decoding result, if occurring two in three kinds of decoding results of every screenshot
Or more identical result, then the decoding result thinks artificial feature recognition result；If only there is a kind of decoding result, in addition two
Kind method can not decode, then taking the decoding result is artificial feature recognition result；
Judge whether artificial feature recognition result is empty: if artificial feature recognition result for sky, enter full figure check step after
It is continuous to execute；If artificial feature recognition result is not sky, obtains commodity recognition result and entrance commodity are checked step and continued to execute；
It is described to judge whether artificial feature recognition result is empty:
If there is following situations, determine artificial feature recognition result for sky: three kinds of methods can not all decode；A kind of decoding knot
Fruit is sky, in addition two kinds of decoding result differences；Three kinds of decoding results are different；
It is described to open operation and closed operation:
The opening and closing operations are the methods of Morphological scale-space, first corrode the operation expanded afterwards and are referred to as out operation；
The closed operation refers to the operation for first expanding post-etching.
8. a kind of product identification system based on artificial feature characterized by comprising
Image capture module: shooting to commodity are obtained in sales counter, and image can all be triggered by opening and closing when cabinet door of selling goods
Acquisition, the picture that Image Acquisition is obtained upload；
Commodity attribute module: according to the picture of upload, identification feature is treated using the feature training pattern of acquisition and is positioned, is obtained
Obtain positioning result；
Commodity identification clearance module: it is identified and is cleared according to the positioning result of acquisition and feature to be identified.
9. the product identification system according to claim 1 based on artificial feature, which is characterized in that further include:
Feature preparation module: making artificial feature is attached on commodity, and commodity are put into sales counter；
Model training module: to the apparent commodity of feature, neural metwork training is carried out to product features, obtains the obvious commodity of feature
Training pattern；For the commodity of artificial feature are added, neural metwork training is carried out to the artificial feature of such commodity, is obtained artificial
Feature commodity training pattern；
The feature training pattern includes: the obvious commodity training pattern of feature, artificial feature commodity training pattern；
Feature to be identified includes: product features, artificial feature；
The commodity identification clearance module, comprising:
Positioning result judgment module: judge whether positioning result is empty: if positioning result for sky, enter full figure check module after
It is continuous to execute, if positioning result is not sky, enters commodity identification module and continue to execute
Commodity identification module: according to the positioning result of acquisition, identifying product features and artificial feature, obtains commodity identification
As a result it obtains commodity recognition result if artificial feature recognition result is sky with artificial feature recognition result and calls full figure multiple
Module is looked into, if artificial feature recognition result is not sky, commodity recognition result is obtained and commodity is called to check module；
Full figure checks module: doing to full figure and turns distortion processing, is then decoded according to industrial decoding standard, decoding result is made
For final commodity recognition result, commodity is called to check module；
Commodity check module: according to the commodity recognition result and artificial feature recognition result of acquisition, will open the door and close the door two
Merchandise news compares in the sales counter at moment, obtains the final result of commodity identification；
The feature preparation module:
The artificial feature includes: two dimensional code, point horizontal and vertical parity check code；
The sales counter includes: sales counter (1), camera (2) and headlamp (3)；
Camera (2), headlamp (3) and the commodity (4) are all in sales counter (1), and the camera is in sales counter
(1) above the commodity (4) in, the artificial feature above feature and part commodity to cargo is shot；
The commodity attribute module:
It according to the picture of upload, is positioned using artificial feature commodity training pattern, obtains all artificial features in picture
Coordinate is confined according to the coordinate pair artificial feature of all artificial features, and carries out figure to the content in frame according to posting
Piece interception, exports positioning result and screenshot；
For no commodity for pasting artificial feature, commodity are positioned using feature obvious commodity training pattern, output is fixed
Position result.
If positioning result is sky, full figure is called to check module, if positioning result is not sky, calls commodity identification module；
The commodity identification module:
The commodity of no patch artificial feature are carried out according to the positioning result of acquisition using the obvious commodity training pattern of feature
Identification obtains commodity recognition result；
For pasting the commodity of artificial feature, the screenshot of acquisition is decoded, comprising:
Operation and closed operation are opened to the artificial feature in screenshot, then according to industrial decoding standard corresponding to the artificial feature
It is decoded, obtains the first decoding result；
To in screenshot artificial feature degree of comparing enhancing, then according to industrial decoding standard corresponding to the artificial feature into
Row decoding, obtains the second decoding result；
Artificial feature in screenshot is carried out to turn distortion correction, then according to industrial decoding standard corresponding to the artificial feature into
Row decoding, obtains the second decoding result；
According to the first, second of acquisition and third decoding result, if occurring two in three kinds of decoding results of every screenshot
Or more identical result, then the decoding result thinks artificial feature recognition result；If only there is a kind of decoding result, in addition two
Kind method can not decode, then taking the decoding result is artificial feature recognition result；
Judge whether artificial feature recognition result is empty: if artificial feature recognition result calls full figure to check module for sky；If
Artificial feature recognition result is not sky, then obtains commodity recognition result and commodity is called to check module；
It is described to judge whether artificial feature recognition result is empty:
If there is following situations, determine artificial feature recognition result for sky: three kinds of methods can not all decode；A kind of decoding knot
Fruit is sky, in addition two kinds of decoding result differences；Three kinds of decoding results are different；
It is described to open operation and closed operation:
The opening and closing operations are the methods of Morphological scale-space, first corrode the operation expanded afterwards and are referred to as out operation；
The closed operation refers to the operation for first expanding post-etching.
10. a kind of computer readable storage medium for being stored with computer program, which is characterized in that the computer program is located
The step of reason device realizes the commodity recognition method described in any one of claims 1 to 7 based on artificial feature when executing.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910549890.4A CN110443118B (en) | 2019-06-24 | 2019-06-24 | Commodity identification method, system and medium based on artificial features |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910549890.4A CN110443118B (en) | 2019-06-24 | 2019-06-24 | Commodity identification method, system and medium based on artificial features |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110443118A true CN110443118A (en) | 2019-11-12 |
| CN110443118B CN110443118B (en) | 2021-09-03 |
Family
ID=68428262
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910549890.4A Active CN110443118B (en) | 2019-06-24 | 2019-06-24 | Commodity identification method, system and medium based on artificial features |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110443118B (en) |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111144417A (en) * | 2019-12-27 | 2020-05-12 | 创新奇智(重庆)科技有限公司 | Intelligent container small target detection method and detection system based on teacher student network |
| CN111209911A (en) * | 2020-01-07 | 2020-05-29 | 创新奇智(合肥)科技有限公司 | Custom tag identification system and identification method based on semantic segmentation network |
| CN111340009A (en) * | 2020-05-15 | 2020-06-26 | 支付宝(杭州)信息技术有限公司 | Identification method and device |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108920992A (en) * | 2018-08-08 | 2018-11-30 | 长沙理工大学 | A kind of positioning and recognition methods of the medical label bar code based on deep learning |
| CN109635705A (en) * | 2018-12-05 | 2019-04-16 | 上海交通大学 | A kind of commodity recognition method and device based on two dimensional code and deep learning |
| CN109816045A (en) * | 2019-02-11 | 2019-05-28 | 青岛海信智能商用***股份有限公司 | A kind of commodity recognition method and device |
| CN109919211A (en) * | 2019-02-26 | 2019-06-21 | 南京旷云科技有限公司 | Commodity recognition method, device, system and computer-readable medium |
-
2019
- 2019-06-24 CN CN201910549890.4A patent/CN110443118B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108920992A (en) * | 2018-08-08 | 2018-11-30 | 长沙理工大学 | A kind of positioning and recognition methods of the medical label bar code based on deep learning |
| CN109635705A (en) * | 2018-12-05 | 2019-04-16 | 上海交通大学 | A kind of commodity recognition method and device based on two dimensional code and deep learning |
| CN109816045A (en) * | 2019-02-11 | 2019-05-28 | 青岛海信智能商用***股份有限公司 | A kind of commodity recognition method and device |
| CN109919211A (en) * | 2019-02-26 | 2019-06-21 | 南京旷云科技有限公司 | Commodity recognition method, device, system and computer-readable medium |
Non-Patent Citations (1)
| Title |
|---|
| 张浩淼: "商品识别结算***的设计与实现", 《计算机软件及计算机应用》 * |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111144417A (en) * | 2019-12-27 | 2020-05-12 | 创新奇智(重庆)科技有限公司 | Intelligent container small target detection method and detection system based on teacher student network |
| CN111209911A (en) * | 2020-01-07 | 2020-05-29 | 创新奇智(合肥)科技有限公司 | Custom tag identification system and identification method based on semantic segmentation network |
| CN111340009A (en) * | 2020-05-15 | 2020-06-26 | 支付宝(杭州)信息技术有限公司 | Identification method and device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110443118B (en) | 2021-09-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110443118A (en) | Commodity recognition method, system and medium based on artificial feature | |
| US10198657B2 (en) | All-weather thermal-image pedestrian detection method | |
| US8238605B2 (en) | Digital video target moving object segmentation method and system | |
| CA2867365C (en) | Method, system and computer storage medium for face detection | |
| CN106845406A (en) | Head and shoulder detection method and device based on multitask concatenated convolutional neutral net | |
| CN107944427B (en) | Dynamic face recognition method and computer readable storage medium | |
| CN105469046B (en) | Based on the cascade vehicle model recognizing method of PCA and SURF features | |
| CN107704797B (en) | Real-time detection method, system and equipment based on pedestrians and vehicles in security video | |
| CN107330360A (en) | A kind of pedestrian's clothing colour recognition, pedestrian retrieval method and device | |
| CN111680690A (en) | Character recognition method and device | |
| CN112417955B (en) | Method and device for processing tour inspection video stream | |
| CN107564032A (en) | A kind of video tracking object segmentation methods based on outward appearance network | |
| Chen et al. | A practical solution for ripe tomato recognition and localisation | |
| CN105631405B (en) | Traffic video intelligent recognition background modeling method based on Multilevel Block | |
| CN109284759A (en) | One kind being based on the magic square color identification method of support vector machines (svm) | |
| CN111767754A (en) | Identification code identification method and device, electronic equipment and storage medium | |
| CN113313688B (en) | Energetic material medicine barrel identification method and system, electronic equipment and storage medium | |
| CN107862314B (en) | Code spraying identification method and device | |
| CN112001336A (en) | Pedestrian boundary crossing alarm method, device, equipment and system | |
| CN110175509A (en) | A kind of round-the-clock eye circumference recognition methods based on cascade super-resolution | |
| Zhang et al. | Cherry picking robot vision recognition system based on OpenCV | |
| Yüce et al. | Improved local binary pattern based action unit detection using morphological and bilateral filters | |
| CN110363251B (en) | SKU image classification method and device, electronic equipment and storage medium | |
| Pang et al. | Low frame rate video target localization and tracking testbed | |
| CN111445267A (en) | Food anti-counterfeiting traceability system and method based on RFID |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |