CN114821797A

CN114821797A - Lip language content identification method and device, storage medium and electronic equipment

Info

Publication number: CN114821797A
Application number: CN202210499614.3A
Authority: CN
Inventors: 杨坤; 孙其功; 杨慧; 马堃
Original assignee: Xi'an Shangtang Intelligent Technology Co ltd
Current assignee: Xi'an Shangtang Intelligent Technology Co ltd
Priority date: 2022-05-09
Filing date: 2022-05-09
Publication date: 2022-07-29

Abstract

The disclosure relates to a lip language content identification method, a device, a storage medium and an electronic device. The method comprises the following steps: acquiring a first visible light image and a first thermal image obtained by shooting a target object at a first moment; performing image fusion on the first visible light image and the first thermal image to obtain a first fused image; and performing lip language content identification on the first fusion image to obtain lip language content corresponding to the target object at the first moment. According to the method, the obtained fused image can simultaneously have visible light information and thermal information through image fusion, the visible light information and the thermal information both comprise effective information for predicting the lip language content, the visible light information comprises mouth shape information, the thermal information comprises direction information, degree information and the like of air suction and air exhalation, and the accuracy of a lip language prediction result obtained by comprehensively considering the information is greatly improved.

Description

Lip language content identification method and device, storage medium and electronic equipment

Technical Field

The present disclosure relates to the field of computer vision technologies, and in particular, to a method and an apparatus for recognizing lip language content, a storage medium, and an electronic device.

Background

In many scenes, the information expressed by the information expression object can be acquired through the lip language, and the intention of the information expression object can be more clearly understood through effective prediction of the content of the lip language. In the related art, the lip language content expressed by the information expression object is usually estimated by analyzing the visible light image and extracting the mouth shape information of the information expression object, but the accuracy of the scheme is low, so that the scheme is difficult to meet the requirement of predicting the lip language content in many scenes, and the ground implementation of intention analysis based on the lip language content or other related applications based on the lip language content is also influenced.

Disclosure of Invention

In order to solve at least one technical problem, the present disclosure provides a lip language content recognition method, device, storage medium, and electronic device.

According to an aspect of the present disclosure, there is provided a lip language content recognition method including: acquiring a first visible light image and a first thermal image obtained by shooting a target object at a first moment; performing image fusion on the first visible light image and the first thermal image to obtain a first fused image; and performing lip language content identification on the first fusion image to obtain lip language content corresponding to the target object at the first moment. Based on the configuration, the visible light image and the thermal image at each moment can be acquired, then the visible light image and the thermal image at the same moment are subjected to image fusion, the obtained fusion image simultaneously has visible light information and thermal information through the image fusion, the visible light information and the thermal information both contain effective information for lip language content prediction, the visible light information contains mouth shape information, the thermal information contains direction information, degree information and the like of air suction and air exhalation, and the accuracy of a lip language prediction result obtained by comprehensively considering the information is greatly improved.

In some possible embodiments, the method further comprises: acquiring a second visible light image and a second thermal image obtained by shooting the target object at a second moment, wherein the second moment is any moment different from the first moment; performing the image fusion on the second visible light image and the second thermal image to obtain a second fused image; performing lip language content identification on the first fusion image to obtain lip language content corresponding to the target object at the first time, including: and respectively carrying out lip language content identification on the first fusion image and the second fusion image to obtain lip language content corresponding to the target object at the first moment. Based on the configuration, a fused image sequence formed by fused images can be obtained, the sequence at least comprises a first fused image and a second fused image, lip language content recognition can be performed on each fused image in the fused image sequence, and the lip language content corresponding to each fused image is considered from the overall perspective, so that the lip language content corresponding to each fused image is globally optimized, the context information can be fully considered from the overall perspective, and the accuracy of lip language content recognition is further improved.

In some possible embodiments, the image fusing the first visible-light image and the first thermal image to obtain a first fused image includes: intercepting the first visible light image based on a target area to obtain a first light area image, wherein the target area is an area containing lip language information; intercepting the first thermal image based on the target area to obtain a first thermal area image; and performing channel-based fusion processing on the first light region image and the first heat region image to obtain the first fusion image. With the above configuration, the visible light information in the region of the first visible light image that is valuable for lip content identification and the thermal information in the region of the first thermal image that is valuable for lip content identification can be fused to obtain a first fused image that includes valid visible light information and valid thermal information, the visible light information in the first fused image includes mouth shape information required for lip content identification, and the thermal information in the first fused image includes breath direction and breath degree information required for lip content identification, thereby significantly improving the accuracy of the lip content obtained after lip content identification based on the first fused image.

In some possible embodiments, the performing a channel-based fusion process on the first light region image and the first thermal region image to obtain the first fused image includes: aligning the first light area image and the first heat area image to obtain a second light area image and a second heat area image, wherein a first position in the second light area image and a second position in the second heat area image corresponding to the first position both correspond to the same position in space, and the first position is any position in the second light area image; and performing channel transverse connection on the second light region image and the second heat region image to obtain the first fusion image. Based on the configuration, the aligned second optical region image and the second thermal region image can be directly connected in the transverse channel direction on the basis of alignment processing, the operation is simple, the speed is high, and the fusion of visible light information and thermal information can be effectively carried out.

In some possible embodiments, the performing lip language content recognition on the first fused image to obtain the lip language content corresponding to the target object at the first time includes: performing feature extraction processing on the first fusion image to obtain first feature information; performing lip language content mapping processing on the first characteristic information to obtain the probability that the lip language content at the first moment is each content word; and determining the content word with the highest probability as the lip language content at the first moment. Based on the configuration, the lip language content at the first time can be determined in a mode of feature extraction and lip language content mapping processing, so that the purpose of comprehensively predicting the lip language content based on the visible light information and the thermal information acquired at the first time is achieved.

In some possible embodiments, the performing feature extraction processing on the first fused image to obtain first feature information includes: performing comprehensive feature extraction on the first fusion image to obtain second feature information, wherein the comprehensive feature extraction comprises at least one of direct feature extraction, feature extraction based on dimension reduction filtering and feature extraction based on context; and obtaining the first characteristic information based on the second characteristic information. Based on the configuration, the information of each position in the first fusion image, the information of the key position related to the lip language content identification or the context information can be extracted, so that the information effective for the lip language content identification in the first characteristic information is gathered, the quality of the first characteristic information is improved, the noise is reduced, and the accuracy of the lip language content identification is improved.

In some possible embodiments, the obtaining the first feature information based on the second feature information includes: and performing the comprehensive feature extraction on the second feature information to obtain the first feature information. Based on the configuration, the information which is effective for lip language content identification in the first characteristic information can be further enriched by executing the comprehensive characteristic extraction for multiple times, the quality of the first characteristic information is further improved, the noise is reduced, and the accuracy of the lip language content identification is improved.

In some possible embodiments, the performing lip language content recognition on the first fusion image and the second fusion image respectively to obtain lip language content corresponding to the target object at the first time includes: respectively carrying out lip language content identification on the first fusion image and the second fusion image to obtain lip language content corresponding to the target object at the first moment and lip language content corresponding to the target object at the second moment; and correcting the lip language content corresponding to the target object at the first moment based on the lip language content corresponding to the target object at the second moment. Based on the configuration, the lip content of the user is continuously optimized based on the context lip content, so that the lip content at each moment is the optimal prediction result from the global perspective, and the accuracy of lip content identification is improved.

In some possible embodiments, the method further comprises: acquiring a visible light image sequence and a thermal image sequence obtained by shooting the target image in a target time period, wherein the visible light image sequence comprises the first visible light image and the second visible light image, and the thermal image sequence comprises the first thermal image and the second thermal image; performing time-based pairing processing on the visible light image sequence and the thermal image sequence to obtain an image pair sequence, wherein each image pair in the image pair sequence comprises a visible light image of the visible light image sequence at a target time and a thermal image of the thermal image sequence at the target time, and the target time is a time corresponding to each image pair; carrying out image fusion on each image pair in the image pair sequence to obtain a fused image sequence; and performing lip language content identification on the fusion image sequence to obtain lip language content corresponding to the target object in the target time period. Based on the configuration, the continuous lip language content recognition can be carried out on the target object, so that the lip language content continuously expressed by the target object can be conveniently determined, and the method is suitable for being applied to a continuous lip language information extraction scene.

In some possible embodiments, the method is implemented by a neural network, and the training method of the neural network includes: acquiring a plurality of sample fusion images and annotation information corresponding to each sample fusion image, wherein each sample fusion image is obtained by carrying out image fusion on a sample visible light image and a corresponding sample thermal image, and the annotation information represents lip language content corresponding to the sample fusion images; inputting each sample fusion image into the neural network to identify lip language content, and obtaining lip language prediction content corresponding to each sample fusion image; and adjusting parameters of the neural network according to the lip language prediction content and the labeling information corresponding to each sample fusion image. Based on the configuration, the neural network can have the capability of accurately identifying lip language content by training the neural network, and characters or words with similar mouth shapes but completely different pronunciations can be accurately identified by extracting the accurate lip language content from the image with the thermal information and the visible light information.

According to a second aspect of the present disclosure, there is provided a lip language content recognition apparatus, the apparatus including:

the image acquisition module is used for acquiring a first visible light image and a first thermal image which are obtained by shooting a target object at a first moment;

the image fusion module is used for carrying out image fusion on the first visible light image and the first thermal image to obtain a first fusion image;

and the lip language content identification module is used for carrying out lip language content identification on the first fusion image to obtain the lip language content corresponding to the target object at the first moment.

In some possible embodiments, the image acquiring module is further configured to acquire a second visible light image and a second thermal image obtained by capturing the target object at a second time, where the second time is any time different from the first time;

the image fusion module is further configured to perform the image fusion on the second visible light image and the second thermal image to obtain a second fusion image;

the lip language content identification module is further configured to perform lip language content identification on the first fusion image and the second fusion image respectively to obtain lip language content corresponding to the target object at the first time.

In some possible embodiments, the image fusion module is configured to perform the following operations:

intercepting the first visible light image based on a target area to obtain a first light area image, wherein the target area is an area containing lip language information;

intercepting the first thermal image based on the target area to obtain a first thermal area image;

and performing channel-based fusion processing on the first light region image and the first heat region image to obtain the first fusion image.

aligning the first light area image and the first heat area image to obtain a second light area image and a second heat area image, wherein a first position in the second light area image and a second position in the second heat area image corresponding to the first position both correspond to the same position in space, and the first position is any position in the second light area image;

and performing channel transverse connection on the second light region image and the second heat region image to obtain the first fusion image.

In some possible embodiments, the lip language content recognition module is configured to perform the following operations:

performing feature extraction processing on the first fusion image to obtain first feature information;

performing lip language content mapping processing on the first characteristic information to obtain the probability that the lip language content at the first moment is each content word;

and determining the content word with the highest probability as the lip language content at the first moment.

performing comprehensive feature extraction on the first fusion image to obtain second feature information, wherein the comprehensive feature extraction comprises at least one of direct feature extraction, feature extraction based on dimension reduction filtering and feature extraction based on context;

and obtaining the first characteristic information based on the second characteristic information.

and performing the comprehensive feature extraction on the second feature information to obtain the first feature information.

respectively carrying out lip language content identification on the first fusion image and the second fusion image to obtain lip language content corresponding to the target object at the first moment and lip language content corresponding to the target object at the second moment;

and correcting the lip language content corresponding to the target object at the first moment based on the lip language content corresponding to the target object at the second moment.

In some possible embodiments, the image acquiring module is configured to acquire, at a target time period, a visible light image sequence and a thermal image sequence obtained by capturing the target image, where the visible light image sequence includes the first visible light image and the second visible light image, and the thermal image sequence includes the first thermal image and the second thermal image; the image fusion module is configured to perform pairing processing on the visible light image sequence and the thermal image sequence based on time to obtain an image pair sequence, where each image pair in the image pair sequence includes a visible light image of the visible light image sequence at a target time and a thermal image of the thermal image sequence at the target time, and the target time is a time corresponding to each image pair; carrying out image fusion on each image pair in the image pair sequence to obtain a fused image sequence; and the lip language content identification module is used for carrying out lip language content identification on the fusion image sequence to obtain the lip language content corresponding to the target object in the target time period.

acquiring a plurality of sample fusion images and annotation information corresponding to each sample fusion image, wherein each sample fusion image is obtained by carrying out image fusion on a sample visible light image and a corresponding sample thermal image, and the annotation information represents lip language content corresponding to the sample fusion images;

inputting each sample fusion image into the neural network to identify lip language content, and obtaining lip language prediction content corresponding to each sample fusion image;

and adjusting parameters of the neural network according to the lip language prediction content and the labeling information corresponding to each sample fusion image.

According to a third aspect of the present disclosure, there is provided an electronic device comprising at least one processor, and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the at least one processor implements the lip language content recognition method according to any one of the first aspect by executing the instructions stored by the memory.

According to a fourth aspect of the present disclosure, there is provided a computer-readable storage medium, in which at least one instruction or at least one program is stored, the at least one instruction or the at least one program being loaded and executed by a processor to implement the lip language content identification method according to any one of the first aspect.

According to a fifth aspect of the present disclosure, there is provided a computer program product comprising a computer program which, when executed by a processor, implements the lip language content recognition method according to any one of the above-mentioned first aspects.

It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure.

Other features and aspects of the present disclosure will become apparent from the following detailed description of exemplary embodiments, which proceeds with reference to the accompanying drawings.

Drawings

In order to more clearly illustrate the embodiments of the present specification or the technical solutions and advantages of the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, it is obvious that the drawings in the following description are only some embodiments of the present specification, and other drawings can be obtained by those skilled in the art without inventive efforts.

FIG. 1 shows a flow diagram of a method of lip language content recognition according to an embodiment of the present disclosure;

FIG. 2 shows a flow diagram of a first fused image acquisition method according to an embodiment of the disclosure;

FIG. 3 shows a schematic view of a target area in one case according to an embodiment of the present disclosure;

FIG. 4 shows a schematic diagram of the effect of an alignment process according to an embodiment of the present disclosure;

FIG. 5 illustrates a block diagram of a neural network for lip language content recognition, in accordance with an embodiment of the present disclosure;

FIG. 6 illustrates a scene diagram of lip language content recognition according to an embodiment of the present disclosure;

fig. 7 illustrates a block diagram of a lip language content recognition apparatus according to an embodiment of the present disclosure;

FIG. 8 shows a block diagram of an electronic device in accordance with an embodiment of the disclosure;

fig. 9 shows a block diagram of another electronic device in accordance with an embodiment of the disclosure.

Detailed Description

The technical solutions in the embodiments of the present disclosure will be clearly and completely described below with reference to the drawings in the embodiments of the present disclosure, and it is obvious that the described embodiments are only a part of the embodiments of the present disclosure, and not all of the embodiments. All other embodiments, which can be obtained by a person skilled in the art without inventive step based on the embodiments in the present description, belong to the protection scope of the present invention.

It should be noted that the terms "first," "second," and the like in the description and claims of the present invention and in the drawings described above are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the invention described herein are capable of operation in sequences other than those illustrated or described herein. Furthermore, the terms "comprises," "comprising," and "having," and any variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, system, article, or server that comprises a list of steps or elements is not necessarily limited to those steps or elements expressly listed, but may include other steps or elements not expressly listed or inherent to such process, method, article, or apparatus.

Various exemplary embodiments, features and aspects of the present disclosure will be described in detail below with reference to the accompanying drawings. In the drawings, like reference numbers can indicate functionally identical or similar elements. While the various aspects of the embodiments are presented in drawings, the drawings are not necessarily drawn to scale unless specifically indicated.

The word "exemplary" is used exclusively herein to mean "serving as an example, embodiment, or illustration. Any embodiment described herein as "exemplary" is not necessarily to be construed as preferred or advantageous over other embodiments.

The term "and/or" herein is merely an association describing an associated object, meaning that three relationships may exist, e.g., a and/or B, may mean: a exists alone, A and B exist simultaneously, and B exists alone. In addition, the term "at least one" herein means any one of a plurality or any combination of at least two of a plurality, for example, including at least one of A, B, C, and may mean including any one or more elements selected from the group consisting of A, B and C.

Furthermore, in the following detailed description, numerous specific details are set forth in order to provide a better understanding of the present disclosure. It will be understood by those skilled in the art that the present disclosure may be practiced without some of these specific details. In some instances, methods, means, elements and circuits that are well known to those skilled in the art have not been described in detail so as not to obscure the present disclosure.

People with hearing loss or hearing disorder can communicate with other people as normal people, and the communication becomes the most expected thing for the people. To achieve this, the related art may recognize the mouth shape of the presenter using the visible light image and convert it into a text form, so that the hearing-impaired person can clearly understand the meaning of the presentation. However, similar mouth shapes may exist in lip language content recognition by using visible light images, but the problem of wrong recognition of characters or words with completely different pronunciations affects user experience. For example, the captured visible light image may be used to process to obtain a lip image sequence, and then the lip content recognition processing may be performed on the lip image sequence. This method obviously has a problem that a character or a word having a similar mouth shape but having a completely different pronunciation cannot be accurately recognized. For another example, at least one frame of 3D lip language image of the current user may be obtained by using a 3D camera, and then a lip language content recognition result is obtained through the 3D lip language content recognition model. But the method still has the problem that the characters or words with similar mouth shapes but completely different pronunciations cannot be accurately recognized.

That is to say, most of lip content recognition devices or systems in the related art only use visible light images or electromyographic signals to perform lip content recognition, and cannot recognize characters or words with similar mouth shapes but with completely different pronunciations. This creates difficulties for landing applications that provide predictive functionality of lip language content and other related applications that are predicted based on lip language content.

Of course, there are also related techniques for lip language content prediction using thermal images. For example, thermal imaging may be used to determine the face region, and then visible light video frames may be used for lip language content recognition. However, the method does not use the thermal image as additional information for lip language content recognition, that is, when lip language prediction is performed, information which may exist in the thermal image and can be used for improving the accuracy of lip language content prediction is still not considered, and therefore, the problem of inaccurate recognition may occur for words with similar mouth shapes but with completely different pronunciations.

The embodiment of the disclosure considers that characters or words with similar mouth shapes but completely different pronunciations have slightly different mouth shapes and different air spitting modes, and the slight difference reflected in the visible light image may be amplified in the thermal imaging picture, so that the accuracy of lip language content identification can be effectively improved by adding the information of the thermal image in the lip language content identification. Based on this consideration, the embodiments of the present disclosure provide a lip language content identification method.

The embodiment of the disclosure provides a lip language content identification method, which can acquire visible light images and thermal images at various moments, then perform image fusion on the visible light images and the thermal images at the same moment, enable the obtained fusion images to simultaneously have visible light information and thermal information through the image fusion, enable the visible light information and the thermal information to both include effective information for lip language content prediction, enable the visible light information to include mouth shape information, and enable the thermal information to include direction information, degree information and the like of air suction and air discharge, and greatly improve the accuracy of a lip language prediction result obtained by comprehensively considering the information.

The lip language content identification method provided by the embodiment of the present disclosure may be executed by a terminal device, a server, or other types of electronic devices, where the terminal device may be a User Equipment (UE), a mobile device, a User terminal, a cellular phone, a cordless phone, a Personal Digital Assistant (PDA), a handheld device, a computing device, an in-vehicle device, a wearable device, or the like. In some possible implementations, the lip language content recognition method may be implemented by a processor calling computer readable instructions stored in a memory. The lip language content recognition method according to the embodiment of the present disclosure will be described below by taking an electronic device as an execution subject.

Fig. 1 is a flowchart illustrating a lip language content recognition method according to an embodiment of the present disclosure, where as shown in fig. 1, the method includes:

s10: a first visible light image and a first thermal image obtained by shooting a target object at a first moment are acquired.

The first time is not limited in the embodiments of the present disclosure, and may be any time at which lip language content recognition is required. And simultaneously shooting a target object through a visible light camera and a thermal camera to obtain the first visible light image and the first thermal image. The type and model of the visible light camera are not limited in the embodiments of the present disclosure. Illustratively, it may be a black and white camera or an rgb (red Green blue) camera. The RGB camera is an image pickup apparatus that receives light reflected from an object by using Red, Green, and Blue filters to generate an RGB color image. The Thermal Camera is also referred to as a Thermal imaging Camera, and the Thermal imaging Camera is an Infrared Camera, which may be a Thermal Camera (Thermal Camera) or an Infrared Camera (Infrared Radiation Camera), for example. The visible light camera and the thermal camera can be cameras located at different positions respectively and can also be integrated together to form a binocular camera.

The target object is not limited in the embodiments of the present disclosure, and may be any object expressed by lip language, such as a human, an animal, or even a smart device simulating a human. Of course, the region where the part of the above-mentioned arbitrary object expressed by lip language is located may be the region where the part expressed by lip language is located, for example, the region may be the lip region of a human, an animal, or even a smart device simulating a human.

S20: and carrying out image fusion on the first visible light image and the first thermal image to obtain a first fused image.

In the embodiment of the present disclosure, the information fusion of the visible light information in the first visible light image and the thermal information in the first thermal image is realized by performing image fusion on the first visible light image and the first thermal image, but the embodiment of the present disclosure does not limit the specific manner of image fusion, and for example, the image fusion may be performed by various manners such as pixel-by-pixel product, weighted sum, convolution, pooling, and the like, as long as the fusion purpose can be achieved.

Fig. 2 is a flowchart illustrating a first fused image acquiring method according to an embodiment of the present disclosure, where as shown in fig. 2, the image fusing the first visible light image and the first thermal image to obtain a first fused image includes:

s21: and performing interception on the first visible light image based on a target area to obtain a first light area image, wherein the target area is an area containing lip language information.

The embodiment of the present disclosure does not limit the specific range of the target area nor the determination manner of the target area, and may be set according to actual situations, for example, as shown in fig. 3, which illustrates the target area in one case. Two rays A and B are emitted vertically downward along the external canthus of the eyes, a ray C passing through the highest point of the nose tip is determined, a ray D passing through the middle point of the line connecting the lowest position of the lower lip contour line and the lowest position of the chin is determined, and the region enclosed by the ray A, B, C, D is the target region. In some cases, some visible light images may also be extracted based on a Region of interest (RoI) through a neural network, and the extraction result is used for lip language prediction, and an optimal range of the RoI is determined according to the accuracy of the prediction, and the optimal range is determined as the target Region.

And cutting out a part belonging to the target area in the first visible light image to obtain the first light area image.

S22: and intercepting the first thermal image based on the target area to obtain a first thermal area image.

Similarly, the portion of the first thermal image that belongs to the target area may be cut out to obtain the first thermal area image.

S23: and performing channel-based fusion processing on the first light region image and the first thermal region image to obtain the first fused image.

The embodiment of the present disclosure is not limited to the way of performing the channel-based fusion processing on the first light region image and the first thermal region image, and for example, the channel-by-channel content addition, multiplication, and the like may be performed as long as the information of each channel in the first light region image and the channel information in the first thermal region image can be embodied in the first fused image.

With the above configuration, the visible light information in the region of the first visible light image that is valuable for lip content recognition and the thermal information in the region of the first thermal image that is valuable for lip content recognition can be fused together to obtain a first fused image that includes valid visible light information and valid thermal information, the visible light information in the first fused image includes mouth shape information required for lip content recognition, and the thermal information in the first fused image includes a breath direction and a breath degree information required for lip content recognition, thereby significantly improving the accuracy of the lip content obtained after lip content recognition based on the first fused image.

In one specific embodiment, the obtaining the first fused image by performing channel-based fusion processing on the first light region image and the first thermal region image includes:

s231: and performing alignment processing on the first light region image and the first thermal region image to obtain a second light region image and a second thermal region image, wherein a first position in the second light region image and a second position in the second thermal region image corresponding to the first position both correspond to the same position in space, and the first position is any position in the second light region image.

Fig. 4 shows a schematic diagram of an alignment process effect according to an embodiment of the present disclosure. The left side of fig. 4 is a second light region image, and the right side of fig. 4 is a second thermal region image, and it can be seen that the upper left corner p1 of the second light region image and the upper left corner p2 of the second thermal region image both correspond to the same point in space, the second light region image and the second thermal region image have the same size, and therefore, for any point 1 located in the second light region image, a point 2 necessarily exists at the same position in the second thermal region image, and the point 1 and the point 2 necessarily correspond to the same point in space.

Before this alignment process is performed, calibration needs to be performed for the first light region image and the first thermal region image. The calibration process is to determine a corresponding relationship between the pixels corresponding to the same point in the space in the first optical region image and the first thermal region image, and the corresponding relationship may be expressed by a Homography Matrix (Homography Matrix). By determining the homography matrix, the first light region image and the first thermal region image can be aligned to obtain a second light region image and a second thermal region image. In the embodiment of the present disclosure, positions of pixel points that represent the same spatial point in different images are associated imaging positions, for example, the first position and the second position may be regarded as associated imaging positions.

Of course, if the aforementioned binocular camera is used to obtain the aforementioned first light region image and the aforementioned first thermal region image, the binocular camera is already calibrated, and therefore, the aforementioned calibration process may be omitted. Illustratively, if the binocular module is composed of a visible light camera a and a thermal camera B, a homography matrix T is obtained through calibration, a dragonfly is shot by using the binocular module, and the position N of the dragonfly in an image output by the thermal camera B can be uniquely determined according to the position M of the dragonfly in the image output by the visible light camera a and the homography matrix T.

S232: and connecting the second light region image and the second heat region image in a channel transverse direction to obtain the first fused image.

For example, if the second light region image has three channels, channel 1, channel 2, and channel 3, and the second heat region image has one channel, channel 4, a first fused image having four channels can be obtained, the first fused image having channel 1, channel 2, channel 3, and channel 4.

Based on the configuration, the aligned second optical region image and the second thermal region image can be directly connected in the transverse channel direction on the basis of alignment processing, the operation is simple, the speed is high, and the fusion of visible light information and thermal information can be effectively carried out.

S30: and performing lip language content identification on the first fusion image to obtain lip language content corresponding to the target object at the first moment.

The embodiments of the present disclosure do not limit the specific method for lip language content recognition, for example, the lip language content recognition may be implemented by various neural networks, and the embodiments of the present disclosure do not limit the neural networks, for example, the lip language content recognition may be implemented by any one or a combination of a convolutional neural network, a deep neural network, a bidirectional cyclic neural network, and a codec based on an attention mechanism.

Based on the above configuration, the accuracy of the lip language prediction result obtained by comprehensively considering the information can be greatly improved by acquiring the visible light image and the thermal image at the first moment, then performing image fusion on the visible light image and the thermal image at the moment, enabling the obtained fusion image to simultaneously have visible light information and thermal information through the image fusion, wherein the visible light information and the thermal information both comprise effective information for lip language content prediction, the visible light information comprises mouth shape information, and the thermal information comprises direction information, degree information and the like for inhaling and exhaling.

In one embodiment, the method further comprises:

s40: and acquiring a second visible light image and a second thermal image obtained by shooting the target object at a second moment, wherein the second moment is any moment different from the first moment.

The second time is not limited in the embodiments of the present disclosure, and may be any time before or after the first time, that is, at any time different from the first time, the corresponding visible light image and thermal image may be obtained, so as to form a time sequence, and the visible light image and thermal image corresponding to each time point in the time sequence are obtained, of course, the time sequence is a time sequence including the first time and the second time.

S50: and performing the image fusion on the second visible light image and the second thermal image to obtain a second fused image.

Step S50 may be understood as performing image fusion on the visible light image sequence and the thermal image corresponding to each time point in the time sequence to obtain a corresponding fused image. The process of obtaining the fusion image may refer to the foregoing, and is not described herein again.

In one scene, a period of continuous shooting can be performed on the target object, and lip language content recognition is performed on the shooting result to obtain lip language content expressed by the target object within the period of time. Specifically, a visible light image sequence and a thermal image sequence obtained by capturing the target image may be acquired at a target time period, where the visible light image sequence includes the first visible light image and the second visible light image, and the thermal image sequence includes the first thermal image and the second thermal image; performing a time-based matching process on the visible light image sequence and the thermal image sequence to obtain a sequence of image pairs, each image pair in the sequence of image pairs including a visible light image of the visible light image sequence at a target time and a thermal image of the thermal image sequence at the target time, the target time being a time corresponding to each image pair; carrying out image fusion on each image pair in the image pair sequence to obtain a fused image sequence; and performing lip language content identification on the fusion image sequence to obtain lip language content corresponding to the target object in the target time period.

It can be understood that the above-mentioned process of performing image fusion on each image pair is based on the same inventive concept as the image fusion in the foregoing, and the method of performing lip language recognition on the fused image sequence is also consistent with the foregoing, and is not described herein again.

In many cases, the target object may be continuously photographed to obtain a video formed by a visible light video and a thermal image, each frame in the video formed by the visible light video and the thermal image may be extracted to obtain a visible light image sequence and a thermal image sequence, or key frames in the video formed by the visible light video and the thermal image may be extracted to obtain a visible light image sequence and a thermal image sequence, or even a representative frame in the video formed by the visible light video and the thermal image may be extracted by a preset adjacent frame redundancy detection algorithm to form a visible light image sequence and a thermal image sequence, where the representative frame may be understood as a frame whose redundancy with the adjacent frame is smaller than a preset threshold. That is, the embodiments of the present disclosure do not limit the method for determining the visible light image sequence and the thermal image sequence from the video formed by the visible light video and the thermal image, and may be reasonably selected according to the actual situation. Based on the configuration, the continuous lip language content recognition can be carried out on the target object, so that the lip language content continuously expressed by the target object can be conveniently determined, and the method is suitable for being applied to a continuous lip language information extraction scene.

In an embodiment, the performing lip language content recognition on the first fused image to obtain lip language content corresponding to the target object at the first time includes: and respectively carrying out lip language content identification on the first fusion image and the second fusion image to obtain lip language content corresponding to the target object at the first moment.

Specifically, lip language content recognition may be performed on the first fused image and the second fused image, respectively, to obtain lip language content corresponding to the target object at the first time and lip language content corresponding to the target object at the second time; and correcting the lip language content corresponding to the target object at the first moment based on the lip language content corresponding to the target object at the second moment. Through the continuous step of optimizing the lip language content of the user based on the context, the lip language content at each moment can be the optimal prediction result from the global perspective, and the accuracy of lip language content identification is improved.

In fact, the foregoing steps may obtain a fused image sequence formed by fused images, lip language content identification may be performed on each fused image in the fused image sequence, and lip language content corresponding to each fused image is considered from an overall perspective, and lip language content corresponding to each fused image is globally optimized, so that context information may be considered from an overall perspective, and accuracy of lip language content identification is further improved again.

In this embodiment, taking a first fused image as an example, performing lip language content recognition on the first fused image to obtain lip language content corresponding to the target object at the first time includes:

s31: and performing feature extraction processing on the first fusion image to obtain first feature information.

The embodiment of the present disclosure does not limit the specific method of the feature extraction process, for example, the feature may be extracted by at least one convolution process, or may be extracted by an attention mechanism. In a possible implementation manner, comprehensive feature extraction may be performed on the first fused image to obtain second feature information, where the comprehensive feature extraction includes at least one of direct feature extraction, feature extraction based on dimension reduction filtering, and feature extraction based on context; the first feature information is obtained based on the second feature information.

The embodiments of the present disclosure do not limit specific methods of direct feature extraction, feature extraction based on dimension reduction filtering, and feature extraction based on context, for example, at least one level of convolution or self-attention extraction belongs to direct feature extraction, operations that can achieve a dimension reduction effect, such as maximum pooling processing, global pooling processing, compression processing, and the like, all belong to feature extraction based on dimension reduction filtering, and in consideration of processing of context, for example, a network, such as a cyclic convolution neural network or a codec, takes the context into consideration when performing data processing, and further, for example, local response normalization processing also belongs to feature extraction based on context. If the above-described feature extraction is performed for each fused image in the sequence of fused image formation including the first fused image, context-based feature extraction may be performed. For example, in a specific embodiment, the convolution operation, the maximum pooling operation, and the local response normalization operation may be performed on the first fused image in sequence to obtain the second feature information.

Based on the configuration, the information of each position in the first fusion image, the information of the key position related to the lip language content identification or the context information can be extracted, so that the information effective for the lip language content identification in the first characteristic information is gathered, the quality of the first characteristic information is improved, the noise is reduced, and the accuracy of the lip language content identification is improved.

In some embodiments, the obtaining the first feature information based on the second feature information includes: and performing the comprehensive feature extraction on the second feature information to obtain the first feature information. That is to say, the second feature information may be extracted again by using the composite feature, of course, the extraction manner or specific extraction parameters used in the two times of composite feature extraction may be different, and the execution times of the composite feature extraction is not limited in the present disclosure, for example, the first fused image may be subjected to feature extraction more than 2 times, and the first feature information is finally obtained. Based on the configuration, the information which is effective for lip language content identification in the first characteristic information can be further enriched by executing the comprehensive characteristic extraction for multiple times, the quality of the first characteristic information is further improved, the noise is reduced, and the accuracy of the lip language content identification is improved.

S32: and performing lip language content mapping processing on the first characteristic information to obtain the probability that the lip language content at the first moment is each content word.

The embodiments of the present disclosure do not limit the method used in the lip language content mapping process, for example, the lip language content mapping may be completed by using various models capable of performing language prediction in a neural network. For example, a convolutional neural network in the neural network may be used to complete mapping of the first feature information to a space where the lip language content is located, and then the probability that the lip language content at the first time is each content word is obtained through the full connection layer and the activation layer. The content word is a word in a space where the lip language content is located, and is a word which can be expressed by lip language in a lip language expression scene.

S33: and determining the content word with the highest probability as the lip language content at the first moment.

Based on the configuration, the lip language content at the first time can be determined in a mode of feature extraction and lip language content mapping processing, so that the purpose of comprehensively predicting the lip language content based on the visible light information and the thermal information acquired at the first time is achieved.

In some embodiments, the lip language content recognition method may be implemented by a neural network, and the training method of the neural network includes:

s1: acquiring a plurality of sample fusion images and annotation information corresponding to each sample fusion image, wherein each sample fusion image is obtained by carrying out image fusion on a sample visible light image and a corresponding sample thermal image, and the annotation information represents lip language content corresponding to the sample fusion images.

The method for obtaining the sample fusion image is the same as the method for obtaining the fusion image, and is not described herein again.

S2: and inputting each sample fusion image into the neural network to identify lip language content, so as to obtain lip language prediction content corresponding to each sample fusion image.

The method for the neural network to identify the lip language content of each sample fusion image is the same as the method for identifying the lip language content in the foregoing, and is not repeated herein.

In one embodiment, please refer to fig. 5, which illustrates a framework diagram of lip language content recognition based on a neural network. The first sample feature is obtained by performing the comprehensive feature extraction at least once. Each comprehensive feature extraction is realized through a convolution layer, a maximum pooling layer and a local response normalization layer, and the convolution layer, the maximum pooling layer and the local response normalization layer are respectively used for sequentially performing direct feature extraction, feature extraction based on dimension reduction filtering and feature extraction based on context on the sample fusion image to obtain a second sample feature. And performing comprehensive feature extraction on the second sample feature at least once again to obtain the first sample feature, and performing lip language content mapping through the convolutional neural network, the full connection layer and the activation layer to obtain lip language prediction content.

And S3, adjusting parameters of the neural network according to the lip language prediction content and the label information corresponding to each sample fusion image.

The present disclosure does not limit the method for adjusting the parameters, for example, a gradient descent method or a least square method may be used to determine the loss generated by the neural network according to the lip language prediction content corresponding to each sample fusion image and the label information, and the parameters of the neural network may be adjusted based on the loss until the condition for stopping adjusting the parameters is reached. The present disclosure does not limit the above condition, and for example, if it is determined that the value of the loss converges to a preset threshold value, or the number of times of adjusting the parameter reaches a preset number of times, the above condition may be considered to be satisfied. After the parameter adjustment is finished, the training is considered to be finished, and the trained neural network can be used for lip language content recognition. Based on the configuration, the neural network can have the capability of accurately identifying lip language content by training the neural network, and characters or words with similar mouth shapes but completely different pronunciations can be accurately identified by extracting the accurate lip language content from the image with the thermal information and the visible light information.

After training is completed, the neural network can be applied to a practical scene to identify lip language content, for example, lip language content corresponding to a target object at a first time and a second time is identified and displayed on a display screen. Referring to fig. 6, which shows a scene schematic diagram of lip language content recognition, a visible light region image and a hot region image can be obtained by simultaneously shooting a target object by using a visible light camera and a hot camera and intercepting an interesting region of a lip in a shooting result, after image fusion is performed on the two images, lip language content recognition is performed on a fusion result, and then a recognition result can be displayed on a display screen. Based on the configuration, many users with hearing loss or hearing loss can know the intention of the expressive person by reading the lip language content on the display screen, so that the communication efficiency is improved.

It will be understood by those skilled in the art that in the method of the present invention, the order of writing of the steps does not imply a strict order of execution and any limitations on the implementation, and the specific order of execution of the steps should be determined by their function and possible inherent logic.

It is understood that the above-mentioned method embodiments of the present disclosure can be combined with each other to form a combined embodiment without departing from the logic of the principle, which is limited by the space, and the detailed description of the present disclosure is omitted.

Fig. 7 illustrates a lip language contents recognition apparatus according to an embodiment of the present disclosure, the apparatus including:

the image acquisition module 10 is configured to acquire a first visible light image and a first thermal image obtained by shooting a target object at a first time;

an image fusion module 20, configured to perform image fusion on the first visible light image and the first thermal image to obtain a first fused image;

a lip content recognition module 30, configured to perform lip content recognition on the first fused image to obtain lip content corresponding to the target object at the first time.

In some possible embodiments, the image acquiring module 10 is further configured to acquire a second visible light image and a second thermal image obtained by capturing the target object at a second time, where the second time is any time different from the first time;

the image fusion module 20 is further configured to perform the image fusion on the second visible light image and the second thermal image to obtain a second fused image;

the lip content recognition module 30 is further configured to perform lip content recognition on the first fused image and the second fused image, respectively, to obtain lip content corresponding to the target object at the first time.

In some possible embodiments, the image fusion module 20 is configured to perform the following operations:

and performing channel-based fusion processing on the first light region image and the first thermal region image to obtain the first fused image.

aligning the first light region image and the first thermal region image to obtain a second light region image and a second thermal region image, wherein a first position in the second light region image and a second position in the second thermal region image corresponding to the first position both correspond to the same position in space, and the first position is any position in the second light region image;

In some possible embodiments, the lip language content recognition module 30 is configured to perform the following operations:

performing comprehensive feature extraction on the first fused image to obtain second feature information, wherein the comprehensive feature extraction comprises at least one of direct feature extraction, feature extraction based on dimension reduction filtering and feature extraction based on context;

the first feature information is obtained based on the second feature information.

respectively performing lip language content identification on the first fusion image and the second fusion image to obtain lip language content corresponding to the target object at the first moment and lip language content corresponding to the target object at the second moment;

In some possible embodiments, the image acquiring module 10 is configured to acquire, at a target time period, a visible light image sequence and a thermal image sequence obtained by capturing the target image, where the visible light image sequence includes the first visible light image and the second visible light image, and the thermal image sequence includes the first thermal image and the second thermal image; the image fusion module 20 is configured to perform time-based pairing processing on the visible light image sequence and the thermal image sequence to obtain an image pair sequence, where each image pair in the image pair sequence includes a visible light image of the visible light image sequence at a target time and a thermal image of the thermal image sequence at the target time, and the target time is a time corresponding to each image pair; carrying out image fusion on each image pair in the image pair sequence to obtain a fused image sequence; the lip content recognition module 30 is configured to perform lip content recognition on the fused image sequence to obtain lip content corresponding to the target object in the target time period.

and adjusting parameters of the neural network according to the lip language prediction content and the label information corresponding to each sample fusion image.

In some embodiments, functions of or modules included in the apparatus provided in the embodiments of the present disclosure may be used to execute the method described in the above method embodiments, and specific implementation thereof may refer to the description of the above method embodiments, and for brevity, will not be described again here.

The embodiment of the present disclosure also provides a computer-readable storage medium, where at least one instruction or at least one program is stored in the computer-readable storage medium, and the at least one instruction or the at least one program is loaded by a processor and executed to implement the method. The computer readable storage medium may be a non-volatile computer readable storage medium.

Embodiments of the present disclosure also provide a computer program product comprising a computer program that, when executed by a processor, implements the above-described method.

An embodiment of the present disclosure further provides an electronic device, including: a processor; a memory for storing processor-executable instructions; wherein the processor is configured as the method.

The electronic device may be provided as a terminal, server, or other form of device.

FIG. 8 shows a block diagram of an electronic device in accordance with an embodiment of the disclosure. For example, the electronic device 800 may be a mobile phone, a computer, a digital broadcast terminal, a messaging device, a game console, a tablet device, a medical device, a fitness device, a personal digital assistant, or the like terminal.

Referring to fig. 8, electronic device 800 may include one or more of the following components: processing component 802, memory 804, power component 806, multimedia component 808, audio component 810, input/output (I/O) interface 812, sensor component 814, and communications component 816.

The processing component 802 generally controls overall operation of the electronic device 800, such as operations associated with display, telephone calls, data communications, camera operations, and recording operations. The processing components 802 may include one or more processors 820 to execute instructions to perform all or a portion of the steps of the methods described above. Further, the processing component 802 can include one or more modules that facilitate interaction between the processing component 802 and other components. For example, the processing component 802 can include a multimedia module to facilitate interaction between the multimedia component 808 and the processing component 802.

The memory 804 is configured to store various types of data to support operations at the electronic device 800. Examples of such data include instructions for any application or method operating on the electronic device 800, contact data, phonebook data, messages, pictures, videos, and so forth. The memory 804 may be implemented by any type or combination of volatile or non-volatile memory devices such as Static Random Access Memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic or optical disks.

The power supply component 806 provides power to the various components of the electronic device 800. The power components 806 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power for the electronic device 800.

The multimedia component 808 includes a screen that provides an output interface between the electronic device 800 and a user as described above. In some embodiments, the screen may include a Liquid Crystal Display (LCD) and a Touch Panel (TP). If the screen includes a touch panel, the screen may be implemented as a touch screen to receive an input signal from a user. The touch panel includes one or more touch sensors to sense touch, slide, and gestures on the touch panel. The touch sensor may not only sense the boundary of the touch or slide action but also detect the duration and pressure associated with the touch or slide operation. In some embodiments, the multimedia component 808 includes a front facing camera and/or a rear facing camera. The front camera and/or the rear camera may receive external multimedia data when the electronic device 800 is in an operation mode, such as a photographing mode or a video mode. Each front camera and rear camera may be a fixed optical lens system or have a focal length and optical zoom capability.

The audio component 810 is configured to output and/or input audio signals. For example, the audio component 810 includes a Microphone (MIC) configured to receive external audio signals when the electronic device 800 is in an operational mode, such as a call mode, a recording mode, and a voice recognition mode. The received audio signals may further be stored in the memory 804 or transmitted via the communication component 816. In some embodiments, audio component 810 also includes a speaker for outputting audio signals.

The I/O interface 812 provides an interface between the processing component 802 and peripheral interface modules, which may be keyboards, click wheels, buttons, etc. These buttons may include, but are not limited to: a home button, a volume button, a start button, and a lock button.

The sensor assembly 814 includes one or more sensors for providing various aspects of state assessment for the electronic device 800. For example, the sensor assembly 814 may detect an open/closed state of the electronic device 800, the relative positioning of components, such as a display and keypad of the electronic device 800, the sensor assembly 814 may also detect a change in position of the electronic device 800 or a component of the electronic device 800, the presence or absence of user contact with the electronic device 800, orientation or acceleration/deceleration of the electronic device 800, and a change in temperature of the electronic device 800. Sensor assembly 814 may include a proximity sensor configured to detect the presence of a nearby object in the absence of any physical contact. The sensor assembly 814 may also include a light sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, the sensor assembly 814 may also include an acceleration sensor, a gyroscope sensor, a magnetic sensor, a pressure sensor, or a temperature sensor.

The communication component 816 is configured to facilitate wired or wireless communication between the electronic device 800 and other devices. The electronic device 800 may access a wireless network based on a communication standard, such as WiFi, 2G, 3G, 4G, 5G, or a combination thereof. In an exemplary embodiment, the communication component 816 receives a broadcast signal or broadcast related information from an external broadcast management system via a broadcast channel. In an exemplary embodiment, the above-mentioned communication component 816 further comprises a Near Field Communication (NFC) module to facilitate short-range communication. For example, the NFC module may be implemented based on Radio Frequency Identification (RFID) technology, infrared data association (IrDA) technology, Ultra Wideband (UWB) technology, Bluetooth (BT) technology, and other technologies.

In an exemplary embodiment, the electronic device 800 may be implemented by one or more Application Specific Integrated Circuits (ASICs), Digital Signal Processors (DSPs), Digital Signal Processing Devices (DSPDs), Programmable Logic Devices (PLDs), Field Programmable Gate Arrays (FPGAs), controllers, micro-controllers, microprocessors or other electronic components for performing the above-described methods.

In an exemplary embodiment, a non-transitory computer-readable storage medium, such as the memory 804, is also provided that includes computer program instructions executable by the processor 820 of the electronic device 800 to perform the above-described methods.

Fig. 9 shows a block diagram of another electronic device in accordance with an embodiment of the disclosure. For example, the electronic device 1900 may be provided as a server. Referring to fig. 9, electronic device 1900 includes a processing component 1922 further including one or more processors and memory resources, represented by memory 1932, for storing instructions, e.g., applications, executable by processing component 1922. The application programs stored in memory 1932 may include one or more modules that each correspond to a set of instructions. Further, the processing component 1922 is configured to execute instructions to perform the above-described method.

The electronic device 1900 may also include a power component 1926 configured to perform power management of the electronic device 1900, a wired or wireless network interface 1950 configured to connect the electronic device 1900 to a network, and an input/output (I/O) interface 1958. The electronic device 1900 may operate based on an operating system, such as Windows Server, MacOSXTM, UnixTM, LinuxTM, FreeBSDTM, or the like, stored in the memory 1932.

In an exemplary embodiment, a non-transitory computer readable storage medium, such as the memory 1932, is also provided that includes computer program instructions executable by the processing component 1922 of the electronic device 1900 to perform the above-described methods.

The present disclosure may be systems, methods, and/or computer program products. The computer program product may include a computer-readable storage medium having computer-readable program instructions embodied thereon for causing a processor to implement various aspects of the present disclosure.

The computer readable storage medium may be a tangible device that can hold and store the instructions for use by the instruction execution device. The computer readable storage medium may be, for example, but not limited to, an electronic memory device, a magnetic memory device, an optical memory device, an electromagnetic memory device, a semiconductor memory device, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of the computer readable storage medium would include the following: a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a Static Random Access Memory (SRAM), a portable compact disc read-only memory (CD-ROM), a Digital Versatile Disc (DVD), a memory stick, a floppy disk, a mechanical coding device, such as punch cards or in-groove projection structures having instructions stored thereon, and any suitable combination of the foregoing. Computer-readable storage media as used herein is not to be construed as transitory signals per se, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through a waveguide or other transmission medium (e.g., optical pulses through a fiber optic cable), or electrical signals transmitted through electrical wires.

The computer-readable program instructions described herein may be downloaded from a computer-readable storage medium to a respective computing/processing device, or to an external computer or external storage device via a network, such as the internet, a local area network, a wide area network, and/or a wireless network. The network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers and/or edge servers. The network adapter card or network interface in each computing/processing device receives computer-readable program instructions from the network and forwards the computer-readable program instructions for storage in a computer-readable storage medium in the respective computing/processing device.

The computer program instructions for carrying out operations of the present disclosure may be assembly instructions, Instruction Set Architecture (ISA) instructions, machine related instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, including an object oriented programming language such as Smalltalk, C + or the like, and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The computer-readable program instructions may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet service provider). In some embodiments, the electronic circuitry that can execute the computer-readable program instructions implements aspects of the present disclosure by utilizing the state information of the computer-readable program instructions to personalize the electronic circuitry, such as a programmable logic circuit, a Field Programmable Gate Array (FPGA), or a Programmable Logic Array (PLA).

Various aspects of the present disclosure are described herein with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the disclosure. It will be understood that each block of the flowchart illustrations and/or block diagrams, and combinations of blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer-readable program instructions.

These computer-readable program instructions may be provided to a processor of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions/acts specified in the flowchart and/or block diagram block or blocks. These computer-readable program instructions may also be stored in a computer-readable storage medium that can direct a computer, programmable data processing apparatus, and/or other devices to function in a particular manner, such that the computer-readable medium storing the instructions comprises an article of manufacture including instructions which implement the function/act specified in the flowchart and/or block diagram block or blocks.

The computer readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other devices to cause a series of operational steps to be performed on the computer, other programmable apparatus or other devices to produce a computer implemented process such that the instructions which execute on the computer, other programmable apparatus or other devices implement the functions/acts specified in the flowchart and/or block diagram block or blocks.

The flowchart and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of instructions, which comprises one or more executable instructions for implementing the specified logical function(s). In some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.

Having described embodiments of the present disclosure, the foregoing description is intended to be exemplary, not exhaustive, and not limited to the disclosed embodiments. Many modifications and variations will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The terms used herein were chosen in order to best explain the principles of the embodiments, the practical application, or technical improvements to the techniques in the marketplace, or to enable others of ordinary skill in the art to understand the embodiments disclosed herein.

Claims

1. A lip language content recognition method, characterized in that the method comprises:

acquiring a first visible light image and a first thermal image obtained by shooting a target object at a first moment;

performing image fusion on the first visible light image and the first thermal image to obtain a first fused image;

and performing lip language content identification on the first fusion image to obtain lip language content corresponding to the target object at the first moment.

2. The method of claim 1, further comprising:

acquiring a second visible light image and a second thermal image obtained by shooting the target object at a second moment, wherein the second moment is any moment different from the first moment;

performing the image fusion on the second visible light image and the second thermal image to obtain a second fused image;

performing lip language content identification on the first fusion image to obtain lip language content corresponding to the target object at the first time, including:

and respectively carrying out lip language content identification on the first fusion image and the second fusion image to obtain lip language content corresponding to the target object at the first moment.

3. The method of claim 1 or 2, wherein said image fusing the first visible light image and the first thermal image resulting in a first fused image comprises:

4. The method of claim 3, wherein performing a channel-based fusion process on the first optical region image and the first thermal region image to obtain the first fused image comprises:

aligning the first light area image and the first thermal area image to obtain a second light area image and a second thermal area image;

performing channel transverse connection on the second light area image and the second heat area image to obtain a first fusion image;

wherein a first position in the second light region image and a second position in the second thermal region image corresponding to the first position both correspond to the same position in space, and the first position is any position in the second light region image.

5. The method according to any one of claims 1 to 4, wherein the performing lip language content recognition on the first fused image to obtain lip language content corresponding to the target object at the first time includes:

6. The method according to claim 5, wherein the performing the feature extraction processing on the first fused image to obtain first feature information includes:

7. The method of claim 6, wherein obtaining the first feature information based on the second feature information comprises:

8. The method according to any one of claims 2 to 7, wherein the performing lip language content recognition on the first fused image and the second fused image respectively to obtain lip language content corresponding to the target object at the first time includes:

9. The method according to any one of claims 2-8, further comprising:

acquiring a visible light image sequence and a thermal image sequence obtained by shooting the target image in a target time period, wherein the visible light image sequence comprises the first visible light image and the second visible light image, and the thermal image sequence comprises the first thermal image and the second thermal image;

performing time-based pairing processing on the visible light image sequence and the thermal image sequence to obtain an image pair sequence, wherein each image pair in the image pair sequence comprises a visible light image of the visible light image sequence at a target time and a thermal image of the thermal image sequence at the target time, and the target time is a time corresponding to each image pair;

carrying out image fusion on each image pair in the image pair sequence to obtain a fused image sequence;

and performing lip language content identification on the fusion image sequence to obtain lip language content corresponding to the target object in the target time period.

10. The method according to claims 1-9, wherein the method is implemented by a neural network, and the training method of the neural network comprises:

11. A lip language contents recognition apparatus, characterized in that the apparatus comprises:

12. A computer-readable storage medium, wherein at least one instruction or at least one program is stored in the computer-readable storage medium, and the at least one instruction or the at least one program is loaded and executed by a processor to implement the lip language content recognition method according to any one of claims 1 to 10.

13. An electronic device comprising at least one processor, and a memory communicatively coupled to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the at least one processor implements the lip language content recognition method according to any one of claims 1 to 10 by executing the instructions stored by the memory.