CN114549345A - Image processing method and system for eliminating reflection interference - Google Patents
Image processing method and system for eliminating reflection interference Download PDFInfo
- Publication number
- CN114549345A CN114549345A CN202210085372.3A CN202210085372A CN114549345A CN 114549345 A CN114549345 A CN 114549345A CN 202210085372 A CN202210085372 A CN 202210085372A CN 114549345 A CN114549345 A CN 114549345A
- Authority
- CN
- China
- Prior art keywords
- target object
- image processing
- image
- images
- moving target
- 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.)
- Pending
Links
- 238000003672 processing method Methods 0.000 title claims abstract description 19
- 238000012545 processing Methods 0.000 claims abstract description 59
- 238000012549 training Methods 0.000 claims abstract description 38
- 238000005286 illumination Methods 0.000 claims abstract description 31
- 230000003068 static effect Effects 0.000 claims abstract description 15
- 238000000034 method Methods 0.000 claims abstract description 11
- 230000002708 enhancing effect Effects 0.000 claims description 8
- 230000008030 elimination Effects 0.000 claims description 6
- 238000003379 elimination reaction Methods 0.000 claims description 6
- 239000011521 glass Substances 0.000 description 4
- 238000007796 conventional method Methods 0.000 description 2
- 230000001678 irradiating effect Effects 0.000 description 2
- 230000003044 adaptive effect Effects 0.000 description 1
- 230000000903 blocking effect Effects 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000007613 environmental effect Effects 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 230000003278 mimic effect Effects 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000005070 sampling Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/70—Denoising; Smoothing
-
- 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
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/90—Dynamic range modification of images or parts thereof
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Image Analysis (AREA)
Abstract
The invention provides an image processing method and system for eliminating reflection interference, wherein the method comprises the following steps: a sample collection step: acquiring a plurality of training samples, wherein each training sample comprises two moving target object images shot in a reflective environment, two enhanced illumination images corresponding to the two moving target images and a static target object image shot in a non-reflective environment; model training: and inputting the collected multiple samples into a GAN image processing model for training to obtain a converged GAN image processing model. By adopting the technical scheme of the invention, the reflection interference in the image shooting process can be effectively reduced.
Description
Technical Field
The invention relates to the technical field of image processing, in particular to an image processing method and an image processing system for eliminating reflection interference.
Background
During image acquisition, when the object to be acquired is located behind a transparent object (typically glass) whose surface is smooth and is prone to reflection, it is easy to form reflective interference for imaging the object itself. When processing a reflection image, the conventional method mainly finds the position and contour of reflection interference and uses the surrounding environment to perform smoothing processing.
The existing conventional method is mainly to position the reflection interference through the characteristics of brightness, contrast or shape of the reflection interference and the surrounding, and then smoothly blend the reflection interference with the surrounding environment, but the characteristics of the reflection interference are not obvious under the condition of relatively complex environment, so that the characteristics are not easy to extract, and the effect of eliminating the reflection interference is poor.
Disclosure of Invention
The invention aims to provide an image processing method and system for eliminating reflection interference.
In an embodiment of the present invention, an image processing method for eliminating reflection interference is provided, which includes:
a sample collection step: acquiring a plurality of training samples, wherein each training sample comprises two moving target object images shot in a reflective environment, two enhanced illumination images corresponding to the two moving target images and a static target object image shot in a non-reflective environment;
model training: and inputting the collected multiple samples into a GAN image processing model for training to obtain a converged GAN image processing model, wherein two moving target object images and two enhanced illumination images which are shot in a reflective environment are used as the input of the GAN image processing model, and a static target object image which is shot in a non-reflective environment is used as the output of the GAN image processing model.
In an embodiment of the present invention, after the step of training the model, the method for processing an image to eliminate reflection interference further includes:
a reflection interference elimination step: inputting the two moving target object images and the enhanced illumination image corresponding to the two moving target images into the trained GAN image processing model to obtain the image of the target object with reflection interference eliminated.
In the embodiment of the present invention, in the sample collection step, the manner of enhancing illumination is as follows: an external direct light source or flash lamp is used to illuminate the target object.
In the embodiment of the invention, in the sample collection step, a video shooting mode is adopted to shoot moving target images, and two frames of images with the largest moving distance of the target object in the shot video are selected as two moving target object images in the training sample.
In the embodiment of the invention, in the shot video, the selection process of two moving target object images comprises the following steps:
selecting a frame of image as a first image, and finding out the position of a target object;
and respectively finding out the positions of the target objects in other frames, calculating the distance between the target objects in other frames and the target object in the first image, and selecting the image with the largest distance as the second image.
In an embodiment of the present invention, an image processing system for eliminating reflection interference is further provided, including:
the system comprises a sample acquisition device, a data acquisition device and a data processing device, wherein the sample acquisition device is used for acquiring a plurality of training samples, and each training sample comprises two moving target object images shot in a reflective environment, two enhanced illumination images corresponding to the two moving target images and a static target object image shot in a non-reflective environment;
and the model training device is used for inputting the collected multiple samples into the GAN image processing model for training to obtain a converged GAN image processing model, wherein two moving target object images and two images with enhanced illumination, which are shot in a reflective environment, are used as the input of the GAN image processing model, and one static target object image, which is shot in a non-reflective environment, is used as the output of the GAN image processing model.
In an embodiment of the present invention, the image processing system for eliminating reflection interference further includes:
and the reflection interference elimination device is used for inputting the two moving target object images and the enhanced illumination image corresponding to the two moving target object images into the trained GAN image processing model to obtain the image of the target object with reflection interference eliminated.
In the embodiment of the invention, when the sample acquisition device acquires a sample, the mode of enhancing illumination is as follows: an external direct light source or flash lamp is used to illuminate the target object.
In the embodiment of the invention, the sample acquisition device shoots moving target images by adopting a video shooting mode, and two frames of images with the largest moving distance of the target object in the shot video are selected as two moving target object images in the training sample.
In the embodiment of the invention, in the shot video, the selection process of two moving target object images comprises the following steps:
selecting a frame of image as a first image, and finding out the position of a target object;
and respectively finding out the positions of the target objects in other frames, calculating the distance between the target objects in other frames and the target object in the first image, and selecting the image with the largest distance as the second image.
Compared with the prior art, the image processing method and the image processing system for eliminating the reflection interference enhance the contrast with the surrounding environment and the reflection interference by enhancing the motion characteristics and the illumination characteristics of the target object, so that the characteristics of the target are more obvious and easier to extract.
Drawings
Fig. 1 is a flowchart of an image processing method for eliminating reflection interference according to an embodiment of the present invention.
FIG. 2 is a schematic environmental view of sample collection according to an embodiment of the present invention.
Fig. 3 is a schematic structural diagram of an image processing system for eliminating reflection interference according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
The following describes the implementation of the present invention in detail with reference to specific embodiments.
In the embodiment of the invention, an image processing method for eliminating reflection interference is provided, wherein a dynamic sample image and a sample image for enhancing illumination in a reflection environment are collected as input, a static image in a non-reflection environment is collected as output to train a GAN image processing model, and thus a convergent GAN image processing model is obtained for reflection elimination processing of a subsequent reflection image.
Specifically, as shown in fig. 1, the image processing method for eliminating the reflection interference includes steps S1-S3. The details will be described below.
Step S1, a sample collection step: the method comprises the steps of collecting a plurality of training samples, wherein each training sample comprises two moving target object images shot in a reflective environment, two enhanced illumination images corresponding to the two moving target images and a static target object image shot in a non-reflective environment.
It should be noted that, in order to enhance the characteristics of the target object itself in the acquired image, such as motion characteristics, illumination characteristics, and the like, the contrast between the target object and the surrounding environment and the reflective interference can be enhanced, so that the characteristics of the target are more obvious and easier to be extracted. Specifically, as the target object moves in the continuous multi-frame images, the target itself includes a motion feature, and the surrounding environment and the reflection interference do not have the motion feature. The target object is directly irradiated by using a flash lamp or other additional direct light source during shooting, so that the illumination characteristic of the target is enhanced. The image of the same target object can be repeatedly acquired in the same environment to serve as a sample, and the image of different target objects can also be acquired in the same environment to serve as a sample.
As shown in fig. 2, in the embodiment of the present invention, in the sample collection step, glass blocking the target object is used as a reflection source. Firstly, when glass is arranged between a target object and a camera for shooting, an image of the target object under the condition of a reflective source can be obtained; then, irradiating the target object by adopting an external direct light source or irradiating the target object by adopting a flash lamp of the camera device, thereby enhancing the illumination of the target object; and finally, removing the glass for shooting to obtain the target object image under the condition of no reflection source.
In the embodiment of the invention, the moving target images are shot in a video shooting mode, and two frames of images with the maximum moving distance of the target object in the shot video are selected as two moving target object images in the training sample.
In the shot video, the selection process of two moving target object images comprises the following steps:
selecting a frame of image as a first image, and finding out the position of a target object;
and respectively finding out the positions of the target objects in other frames, calculating the distance between the target objects in other frames and the target object in the first image, and selecting the image with the largest distance as the second image.
Step S2, model training step: and inputting the collected multiple samples into a GAN image processing model for training to obtain a converged GAN image processing model, wherein two moving target object images and two enhanced illumination images which are shot in a reflective environment are used as the input of the GAN image processing model, and a static target object image which is shot in a non-reflective environment is used as the output of the GAN image processing model.
It should be noted that GAN (generic adaptive Network, Generative countermeasure Network) is a machine learning Network, and is composed of a Generative Network and a discriminant Network. The generation network takes as input a random sampling from the underlying space, and its output needs to mimic as much as possible the real samples in the training set. The input of the discrimination network is the real sample or the output of the generation network, and the purpose is to distinguish the output of the generation network from the real sample as much as possible. The relationship between the input samples and the output samples can be obtained by inputting a large number of samples for training. In the embodiment of the invention, two target object images in a motion state and one target object image for enhancing illumination in a static state when a reflection source exists are used as the input of the GAN image processing model, and one target object image in a static state when no reflection source exists is used as the output of the GAN image processing model to train the GAN image processing model, so that the convergent GAN image processing model is obtained.
Step S3, reflection interference elimination step: inputting the two moving target object images and the enhanced illumination image corresponding to the two moving target images into the trained GAN image processing model to obtain the image of the target object with reflection interference eliminated.
It should be noted that after the trained GAN image processing model is obtained, two captured moving target object images and enhanced illumination image images corresponding to the two moving target images are input according to the input requirement of the model, and the GAN image processing model can output the image of the target object without reflection interference.
As shown in fig. 3, in comparison with the image processing method for eliminating reflection interference, in the embodiment of the present invention, an image processing system for eliminating reflection interference is further provided, and includes a sample collecting device 1, a model training device 2, and a reflection interference eliminating device 3.
The sample acquisition device 1 is configured to acquire a plurality of training samples, where each training sample includes two moving target object images captured in a reflective environment, two enhanced illumination images corresponding to the two moving target images, and a still target object image captured in a non-reflective environment.
The model training device 2 is configured to input the collected multiple samples into a GAN image processing model for training, so as to obtain a converged GAN image processing model, where two moving target object images and two enhanced illumination images captured in a reflective environment are used as input of the GAN image processing model, and a static target object image captured in a non-reflective environment is used as output of the GAN image processing model.
And the reflection interference eliminating device 3 is used for inputting the two moving target object images and the enhanced illumination image corresponding to the two moving target images into the trained GAN image processing model to obtain an image of the target object with reflection interference eliminated.
It should be noted that the image processing system for eliminating the reflection interference and the image processing method for eliminating the reflection interference are based on the same concept, and the operation process thereof is described in detail in the image processing method for eliminating the reflection interference, and is not described herein again.
In summary, by using the image processing method for eliminating the reflection interference of the present invention, the contrast with the surrounding environment and the reflection interference is enhanced by enhancing the characteristics of the target object itself, such as the motion characteristics and the illumination characteristics, instead of reducing the contrast between the reflection interference and the surrounding environment, so that the characteristics of the target are more obvious and easier to be extracted.
The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents and improvements made within the spirit and principle of the present invention are intended to be included within the scope of the present invention.
Claims (10)
1. An image processing method for eliminating reflection interference is characterized by comprising the following steps:
a sample collection step: collecting a plurality of training samples, wherein each training sample comprises two moving target object images shot in a reflective environment, two enhanced illumination images corresponding to the two moving target images, and a static target object image shot in a non-reflective environment;
model training: and inputting the collected multiple samples into a GAN image processing model for training to obtain a converged GAN image processing model, wherein two moving target object images and two enhanced illumination images which are shot in a reflective environment are used as the input of the GAN image processing model, and a static target object image which is shot in a non-reflective environment is used as the output of the GAN image processing model.
2. The image processing method for eliminating the reflection interference according to claim 1, wherein after the model training step, the method further comprises:
a reflection interference elimination step: inputting the two moving target object images and the enhanced illumination image corresponding to the two moving target images into the trained GAN image processing model to obtain the image of the target object with reflection interference eliminated.
3. The image processing method for eliminating the reflective interference according to claim 1, wherein in the sample collection step, the manner of enhancing the illumination is: an external direct light source or flash lamp is used to illuminate the target object.
4. The image processing method for eliminating interference of reflections according to claim 1, wherein in the sample collection step, a video capture mode is used to capture moving target images, and two frames of images with the largest moving distance of the target object in the captured video are selected as two moving target object images in the training sample.
5. The image processing method for eliminating the interference of reflected light according to claim 4, wherein the process of selecting two moving target object images in the captured video comprises:
selecting a frame of image as a first image, and finding out the position of a target object;
and respectively finding out the positions of the target objects in other frames, calculating the distance between the target objects in other frames and the target object in the first image, and selecting the image with the largest distance as the second image.
6. An image processing system for eliminating reflection interference, comprising:
the system comprises a sample acquisition device, a data acquisition device and a data processing device, wherein the sample acquisition device is used for acquiring a plurality of training samples, and each training sample comprises two moving target object images shot in a reflective environment, two enhanced illumination images corresponding to the two moving target images and a static target object image shot in a non-reflective environment;
and the model training device is used for inputting the collected multiple samples into the GAN image processing model for training to obtain a converged GAN image processing model, wherein two moving target object images and two images with enhanced illumination, which are shot in a reflective environment, are used as the input of the GAN image processing model, and one static target object image, which is shot in a non-reflective environment, is used as the output of the GAN image processing model.
7. The image processing system for eliminating interference of reflections according to claim 6, further comprising:
and the reflection interference elimination device is used for inputting the two moving target object images and the enhanced illumination image corresponding to the two moving target object images into the trained GAN image processing model to obtain the image of the target object with reflection interference eliminated.
8. The image processing system for eliminating the reflective interference according to claim 6, wherein when the sample collection device collects the sample, the illumination is enhanced by: an external direct light source or flash lamp is used to illuminate the target object.
9. The image processing system for eliminating interference of reflected light according to claim 6, wherein the sample acquiring device captures moving target images by video capture, and selects two frames of images with the largest moving distance of the target object in the captured video as two moving target object images in the training sample.
10. The image processing system for eliminating interference of reflections according to claim 9, wherein the selection process of two moving target object images in the captured video comprises:
selecting a frame of image as a first image, and finding out the position of a target object;
and respectively finding out the positions of the target objects in other frames, calculating the distance between the target objects in other frames and the target object in the first image, and selecting the image with the largest distance as the second image.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210085372.3A CN114549345A (en) | 2022-01-25 | 2022-01-25 | Image processing method and system for eliminating reflection interference |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210085372.3A CN114549345A (en) | 2022-01-25 | 2022-01-25 | Image processing method and system for eliminating reflection interference |
Publications (1)
Publication Number | Publication Date |
---|---|
CN114549345A true CN114549345A (en) | 2022-05-27 |
Family
ID=81672084
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210085372.3A Pending CN114549345A (en) | 2022-01-25 | 2022-01-25 | Image processing method and system for eliminating reflection interference |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114549345A (en) |
-
2022
- 2022-01-25 CN CN202210085372.3A patent/CN114549345A/en active Pending
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN113052210B (en) | Rapid low-light target detection method based on convolutional neural network | |
CN105930822A (en) | Human face snapshot method and system | |
CN108038452B (en) | Household appliance gesture rapid detection and identification method based on local image enhancement | |
KR102220174B1 (en) | Learning-data enhancement device for machine learning model and method for learning-data enhancement | |
Perfetti et al. | Cellular neural networks with virtual template expansion for retinal vessel segmentation | |
KR101303877B1 (en) | Method and apparatus for serving prefer color conversion of skin color applying face detection and skin area detection | |
CN106548185B (en) | A kind of foreground area determines method and apparatus | |
CN109446982B (en) | AR glasses-based electric power cabinet pressing plate state identification method and system | |
CN111327835B (en) | Multi-line time-sharing exposure processing method and system for camera | |
CN107609475B (en) | Pedestrian detection false detection extraction method based on light field camera | |
CN109343701A (en) | A kind of intelligent human-machine interaction method based on dynamic hand gesture recognition | |
US20220345612A1 (en) | Camera exposure processing method and system | |
CN101930606A (en) | Field depth extending method for image edge detection | |
CN114821440B (en) | Mobile video stream content identification and analysis method based on deep learning | |
CN112365578A (en) | Three-dimensional human body model reconstruction system and method based on double cameras | |
CN115484371A (en) | Image acquisition method, image acquisition device and readable storage medium | |
CN114022823A (en) | Shielding-driven pedestrian re-identification method and system and storable medium | |
CN111830039A (en) | Intelligent product quality detection method and device | |
CN110827375B (en) | Infrared image true color coloring method and system based on low-light-level image | |
CN106846398B (en) | Image line data scanning-based starry sky target maximum light intensity position identification method | |
CN114549345A (en) | Image processing method and system for eliminating reflection interference | |
KR20220047749A (en) | How to create a slab/finger foreground mask | |
CN111738264A (en) | Intelligent acquisition method for data of display panel of machine room equipment | |
CN104239950A (en) | Piston ring counting system based on machine vision and counting method of piston ring counting system based on machine vision | |
CN114913086A (en) | Face image quality enhancement method based on generation countermeasure network |
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 |