Image processing method and system for eliminating reflection interference

Abstract

The present invention provides an image processing method and system for eliminating reflective interference. The method comprises: a sample collection step: collecting multiple training samples, each training sample comprising two images of a moving target object captured in a reflective environment, two images corresponding to the two moving target images with enhanced illumination, and one image of a stationary target object captured in a non-reflective environment; and a model training step: inputting the collected multiple samples into a GAN image processing model for training to obtain a converged GAN image processing model. The technical solution of the present invention can effectively reduce reflective interference during image capture.

Description

Technical Field

[0001] The present invention relates to the field of image processing technology, and in particular to an image processing method and system for eliminating reflection interference. Background Art

[0002] During image acquisition, when the target is behind a smooth, reflective, transparent object (typically glass), this can easily interfere with the image of the target itself. Conventional methods for processing reflective images primarily identify the location and outline of the reflective interference and then use the surrounding environment to smooth it out.

[0003] The existing conventional method mainly locates the reflective interference by first analyzing its brightness, contrast with the surroundings, or shape, and then smoothly integrates it with the surrounding environment. However, in fact, these features of reflective interference are not obvious in complex environments, making them difficult to extract, resulting in poor results in eliminating reflective interference. Summary of the Invention

[0004] The object of the present invention is to provide an image processing method and system for eliminating reflection interference.

[0005] In an embodiment of the present invention, an image processing method for eliminating reflection interference is provided, which includes:

[0006] Sample collection step: collecting multiple training samples, each training sample including two moving target object images taken in a reflective environment, two enhanced illumination images corresponding to the two moving target images, and one stationary target object image taken in a non-reflective environment;

[0007] Model training steps: Input multiple collected samples into the GAN image processing model for training to obtain a converged GAN image processing model. Among them, two moving target object images and two enhanced lighting images taken in a reflective environment are used as inputs of the GAN image processing model, and one static target object image taken in a non-reflective environment is used as the output of the GAN image processing model.

[0008] In an embodiment of the present invention, the image processing method for eliminating reflective interference further includes, after the model training step:

[0009] Reflective interference elimination step: inputting two moving target object images and enhanced illumination images corresponding to the two moving target images into the trained GAN image processing model to obtain an image of the target object with reflected interference eliminated.

[0010] In the embodiment of the present invention, in the sample collection step, the method of enhancing illumination is: using an external direct light source or a flashlight to illuminate the target object.

[0011] In the embodiment of the present invention, in the sample collection step, a video shooting method is used to shoot the moving target image, and in the shot video, two frames of images with the largest target object movement distance are selected as two moving target object images in the training samples.

[0012] In an embodiment of the present invention, the process of selecting two moving target object images in a captured video includes:

[0013] Select a frame as the first image and find the location of the target object;

[0014] The positions of the target objects in the other frames are found respectively, and the distances between the target objects in the other frames and the target object in the first image are calculated, and the frame image with the largest distance is selected as the second image.

[0015] In an embodiment of the present invention, an image processing system for eliminating reflection interference is further provided, comprising:

[0016] A sample collection device for collecting a plurality of training samples, each training sample comprising two images of a moving target object taken in a reflective environment, two images with enhanced illumination corresponding to the two moving target images, and one image of a stationary target object taken in a non-reflective environment;

[0017] The model training device is used to input the collected multiple samples into the GAN image processing model for training to obtain a converged GAN image processing model, wherein two images of moving target objects and two images with enhanced lighting taken in a reflective environment are used as inputs of the GAN image processing model, and one image of a stationary target object taken in a non-reflective environment is used as output of the GAN image processing model.

[0018] In an embodiment of the present invention, the image processing system for eliminating reflection interference further includes:

[0019] The device for eliminating reflection interference is used to input two moving target object images and enhanced illumination images 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.

[0020] In the embodiment of the present invention, when the sample collection device collects samples, the method of enhancing illumination is: using an external direct light source or a flashlight to illuminate the target object.

[0021] In an embodiment of the present invention, the sample acquisition device captures moving target images by video capture, and selects two frames of images in which the target object moves the largest distance from the captured video as two moving target object images in training samples.

[0022] In an embodiment of the present invention, the process of selecting two moving target object images in a captured video includes:

[0023] Select a frame as the first image and find the location of the target object;

[0024] The positions of the target objects in the other frames are found respectively, and the distances between the target objects in the other frames and the target object in the first image are calculated, and the frame image with the largest distance is selected as the second image.

[0025] Compared with the existing technology, the image processing method and system for eliminating reflective interference of the present invention enhances the contrast between the target object and the surrounding environment and reflective interference by enhancing the motion characteristics and lighting characteristics of the target object itself, making the target features more obvious and easier to extract. By training the GAN image processing model with samples with the above characteristics, the GAN image processing model's processing ability for reflective images can be improved, thereby improving the effect of eliminating reflective interference. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] Figure 1 4 is a flowchart of an image processing method for eliminating reflection interference according to an embodiment of the present invention.

[0027] Figure 2 FIG. 4 is a schematic diagram of an environment for collecting samples according to an embodiment of the present invention.

[0028] Figure 3 2 is a schematic structural diagram of an image processing system for eliminating reflection interference according to an embodiment of the present invention. DETAILED DESCRIPTION

[0029] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0030] The implementation of the present invention is described in detail below with reference to specific embodiments.

[0031] In an embodiment of the present invention, an image processing method for eliminating reflective interference is provided. The method collects dynamic sample images in a reflective environment and sample images with enhanced illumination as input, and collects static images in a non-reflective environment as output to train a GAN image processing model, thereby obtaining a converged GAN image processing model for use in subsequent reflective image elimination processing.

[0032] Specifically, if Figure 1 As shown in FIG, the image processing method for eliminating reflection interference includes steps S1 to S3, which will be described in detail below.

[0033] Step S1, sample collection step: collect multiple training samples, each training sample includes two moving target object images taken in a reflective environment, two enhanced illumination images corresponding to the two moving target images, and one stationary target object image taken in a non-reflective environment.

[0034] It should be noted that in order to enhance the characteristics of the target object itself in the captured image, such as motion characteristics, lighting characteristics, etc., the contrast between the target object and the surrounding environment and reflective interference can be enhanced, so that the characteristics of the target are more obvious and easier to extract. Specifically, as the target object moves in multiple consecutive frames of images, the target itself will contain motion characteristics, while the surrounding environment and reflective interference will not have motion characteristics. Use the flash or other additional direct light source during shooting to directly illuminate the target object and enhance the lighting characteristics of the target itself. Images of the same target object can be repeatedly collected in the same environment as samples, or images of different target objects can be collected in the same environment as samples.

[0035] like Figure 2 As shown, in the sample collection step, glass blocking the target object is used as a reflective light source. First, the glass is placed between the target object and the camera for shooting, thereby obtaining an image of the target object with the reflective light source. Then, the target object is illuminated by an external direct light source or a flash of a camera device, thereby enhancing the illumination of the target object. Finally, the glass is removed and shooting is performed, thereby obtaining an image of the target object without the reflective light source.

[0036] Since it is necessary to capture images during motion, in an embodiment of the present invention, video shooting is used to shoot images of moving targets, and in the shot video, two frames of images with the largest moving distance of the target object are selected as two moving target object images in the training samples.

[0037] The process of selecting two moving target object images in the captured video includes:

[0038] Select a frame as the first image and find the location of the target object;

[0039] The positions of the target objects in the other frames are found respectively, and the distances between the target objects in the other frames and the target object in the first image are calculated, and the frame image with the largest distance is selected as the second image.

[0040] Step S2, model training step: input 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 enhanced lighting images taken in a reflective environment are used as inputs of the GAN image processing model, and one stationary target object image taken in a non-reflective environment is used as output of the GAN image processing model.

[0041] It should be noted that a GAN (Generative Adversarial Network) is a machine learning network consisting of a generator and a discriminator. The generator takes random samples from the latent space as input, and its output is designed to closely mimic real samples in the training set. The discriminator takes real samples or the output of the generator as input, and its goal is to distinguish the generator output from real samples as much as possible. By training with a large number of samples, the relationship between input and output samples can be determined.

[0042] Step S3, reflective interference elimination step: input the two moving target object images and the enhanced illumination images corresponding to the two moving target images into the trained GAN image processing model to obtain an image of the target object with reflective interference eliminated.

[0043] It should be noted that after obtaining the trained GAN image processing model, according to the input requirements of the model, two captured images of the moving target object and the enhanced lighting images corresponding to the two moving target images are input, and the GAN image processing model can output the image of the target object with the reflection interference eliminated.

[0044] like Figure 3 As shown, compared with the above-mentioned image processing method for eliminating reflection interference, an embodiment of the present invention further provides an image processing system for eliminating reflection interference, which includes a sample collection device 1, a model training device 2 and a reflection interference elimination device 3.

[0045] The sample acquisition device 1 is used to collect multiple training samples, each training sample includes two moving target object images taken in a reflective environment, two enhanced lighting images corresponding to the two moving target images, and one stationary target object image taken in a non-reflective environment.

[0046] The model training device 2 is used to input the collected multiple samples into the GAN image processing model for training to obtain a converged GAN image processing model, wherein two images of moving target objects and two images with enhanced lighting taken in a reflective environment are used as inputs of the GAN image processing model, and one image of a stationary target object taken in a non-reflective environment is used as output of the GAN image processing model.

[0047] The reflective interference elimination device 3 is used to input two moving target object images and the enhanced illumination images corresponding to the two moving target images into the trained GAN image processing model to obtain an image of the target object with the reflective interference eliminated.

[0048] It should be noted that the image processing system for eliminating reflective interference and the above-mentioned image processing method for eliminating reflective interference are based on the same concept, and its operation process has been described in detail in the above-mentioned image processing method for eliminating reflective interference, and will not be repeated here.

[0049] In summary, the image processing method for eliminating reflective interference of the present invention enhances the contrast between the target object and the surrounding environment and reflective interference by enhancing the characteristics of the target object itself, such as motion characteristics and lighting characteristics, rather than reducing the contrast between the reflective interference and the surrounding environment. This makes the target features more obvious and easier to extract. The GAN image processing model is trained by samples with the above characteristics, thereby improving the GAN image processing model's processing ability for reflective images.

[0050] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. An image processing method for eliminating reflection interference, characterized in that: include: Sample collection step: collecting multiple training samples, each training sample including two images of a moving target object taken in a reflective environment, two images corresponding to the two moving target object images with enhanced lighting, and one image of a stationary target object taken in a non-reflective environment. The target object may contain motion characteristics, while the surrounding environment and reflective interference do not have motion characteristics. The lighting enhancement method is: using an external direct light source or a flash to illuminate the target object to enhance the lighting characteristics of the target object; Model training steps: Input multiple collected samples into the GAN image processing model for training to obtain a converged GAN image processing model. Among them, two moving target object images and two enhanced lighting images taken in a reflective environment are used as inputs of the GAN image processing model, and one static target object image taken in a non-reflective environment is used as the output of the GAN image processing model.

2. The image processing method for eliminating reflection interference according to claim 1, characterized in that: After the model training step, the following steps are also included: The step of eliminating reflection interference is as follows: inputting two moving target object images and enhanced illumination images corresponding to the two moving target object images into the converged GAN image processing model to obtain an image of the target object with reflection interference eliminated.

3. The image processing method for eliminating reflection interference according to claim 1, wherein: In the sample collection step, a video shooting method is used to shoot the moving target object image, and in the shot video, two frames of images with the largest target object movement distance are selected as two moving target object images in the training samples.

4. The image processing method for eliminating reflection interference according to claim 3, wherein: The process of selecting two moving target object images in the captured video includes: Select a frame as the first image and find the location of the target object; The positions of the target objects in the other frames are found respectively, and the distances between the target objects in the other frames and the target object in the first image are calculated, and the frame image with the largest distance is selected as the second image.

5. An image processing system for eliminating reflection interference, characterized in that: include: A sample collection device for collecting a plurality of training samples, each training sample comprising two images of a moving target object taken in a reflective environment, two images corresponding to the two moving target object images with enhanced illumination, and one image of a stationary target object taken in a non-reflective environment, wherein the target object has motion characteristics, while the surrounding environment and reflective interference do not have motion characteristics, and the illumination is enhanced by illuminating the target object with an external direct light source or a flash to enhance the illumination characteristics of the target object; The model training device is used to input the collected multiple samples into the GAN image processing model for training to obtain a converged GAN image processing model, wherein two images of moving target objects and two images with enhanced lighting taken in a reflective environment are used as inputs of the GAN image processing model, and one image of a stationary target object taken in a non-reflective environment is used as output of the GAN image processing model.

6. The image processing system for eliminating reflection interference according to claim 5, characterized in that: Also includes: The device for eliminating reflection interference is used to input two moving target object images and enhanced illumination images corresponding to the two moving target object images into the converged GAN image processing model to obtain an image of the target object with reflection interference eliminated.

7. The image processing system for eliminating reflection interference according to claim 5, characterized in that: The sample acquisition device captures images of moving target objects by video capture, and selects two frames of images with the largest moving distance of the target objects from the captured video as two images of the moving target objects in training samples.

8. The image processing system for eliminating reflection interference according to claim 7, wherein: The process of selecting two moving target object images in the captured video includes: Select a frame as the first image and find the location of the target object; The positions of the target objects in the other frames are found respectively, and the distances between the target objects in the other frames and the target object in the first image are calculated, and the frame image with the largest distance is selected as the second image.

Images