CN114295108A - Distance measurement method and system for external equipment and infrared telescope - Google Patents
Distance measurement method and system for external equipment and infrared telescope Download PDFInfo
- Publication number
- CN114295108A CN114295108A CN202111554027.1A CN202111554027A CN114295108A CN 114295108 A CN114295108 A CN 114295108A CN 202111554027 A CN202111554027 A CN 202111554027A CN 114295108 A CN114295108 A CN 114295108A
- Authority
- CN
- China
- Prior art keywords
- infrared
- target
- telescope
- image
- detection model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000691 measurement method Methods 0.000 title description 3
- 238000001514 detection method Methods 0.000 claims abstract description 134
- 238000012549 training Methods 0.000 claims abstract description 76
- 238000000034 method Methods 0.000 claims abstract description 61
- 239000011521 glass Substances 0.000 claims abstract description 28
- 238000003331 infrared imaging Methods 0.000 claims abstract description 16
- 238000012545 processing Methods 0.000 claims description 23
- 230000008569 process Effects 0.000 claims description 16
- 210000001525 retina Anatomy 0.000 claims description 15
- 238000013480 data collection Methods 0.000 claims description 8
- 230000008676 import Effects 0.000 claims description 8
- 238000005259 measurement Methods 0.000 claims description 6
- 230000003416 augmentation Effects 0.000 claims description 4
- 230000003190 augmentative effect Effects 0.000 claims description 4
- 230000007246 mechanism Effects 0.000 claims description 3
- 238000002372 labelling Methods 0.000 abstract description 3
- 238000012937 correction Methods 0.000 description 6
- 230000004044 response Effects 0.000 description 4
- 238000006243 chemical reaction Methods 0.000 description 3
- 238000013527 convolutional neural network Methods 0.000 description 3
- 238000013135 deep learning Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 241000282326 Felis catus Species 0.000 description 2
- 238000004364 calculation method Methods 0.000 description 2
- 230000008859 change Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000000605 extraction Methods 0.000 description 2
- 238000011176 pooling Methods 0.000 description 2
- 238000003384 imaging method Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000005057 refrigeration Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Abstract
The invention relates to a distance measuring method and a distance measuring system based on external equipment and an infrared telescope, infrared glass lenses with different curvatures and different refractive indexes are arranged on the same axis in the infrared telescope to form an infrared imaging lens, an infrared photosensitive unit is arranged into an uncooled infrared focal plane array to form an infrared detector in the infrared telescope, and the distance measuring method of the infrared telescope comprises the following steps: determining the category of a target to be detected, calibrating a data set, training a detection model, importing the model, detecting the target, measuring the distance of the target, identifying an image and observing and displaying. The method comprises the steps of acquiring an infrared data set, calibrating the data set and training a model to obtain a detection model for target detection, and performing category identification, target ranging and image identification on an infrared target by using the detection model and then performing identification superposition to complete real-time ranging and labeling of the infrared image target.
Description
Technical Field
The invention relates to a distance measuring system, in particular to a distance measuring method and system based on external equipment and an infrared telescope.
Background
An infrared camera is an electronic device that receives infrared light emitted or reflected by a scene itself to the purpose of imaging the scene. The infrared telescope is an observation product which integrates a miniaturized infrared camera, an electronic display screen and a battery module and can transmit an instant infrared picture to a user. The infrared image can distinguish targets with different temperatures, and is not influenced by ambient light, so the infrared image is commonly used in the fields of field hunting, police reconnaissance and the like, and is particularly applied at night. When the user needs to obtain the distance value of the observed target, two modes of physical ranging and image ranging can be used. The physical distance measurement is to integrate an optical distance measurement module on the infrared telescope, and obtain the distance of a target by calculating the time difference between a transmitted beam and a reflected beam, and the physical distance measurement can increase the volume, the weight, the power consumption and the cost of the infrared telescope, so the image distance measurement has more advantages; the image ranging is a method for directly obtaining a target distance by using an infrared image without other equipment, and common image ranging methods are classified into a manual measuring method and a grid contrast method. The manual measuring method is that the upper and lower boundaries of a target are manually marked on a screen, the type or physical height value of the target is input, and the system can obtain the distance value of the target through calculation; the grid comparison method is that horizontal or vertical scales are superposed on an electronic display screen, after the type of a target is selected, the size of scale values occupied by the target at different distances is displayed on the screen, and the estimated value of the target distance can be obtained by comparing the actual scale values occupied by the target with theoretical values.
The grid comparison method cannot accurately obtain the distance value of the target, and the error of distance estimation is large; the manual measurement method is complex to operate, lacks real-time performance, and is particularly difficult to measure the distance of a moving target.
Disclosure of Invention
The technical problem solved by the invention is as follows: an infrared telescope ranging method and system are constructed, and the technical problems of complex operation and large ranging error of the infrared telescope ranging in the prior art are solved.
The technical scheme of the invention is as follows: the method for measuring the distance of the infrared telescope is constructed, infrared glass lenses with different curvatures and different refractive indexes are arranged on the same axis in the infrared telescope to form an infrared imaging lens, an infrared photosensitive unit is arranged into an uncooled infrared focal plane array to form an infrared detector in the infrared telescope, and the method is characterized by comprising the following steps of:
model import: aiming at the used infrared telescope embedded hardware platform, after converting the training detection model into a format matched with the embedded platform, the embedded system of the infrared telescope is led into the training detection model; the training detection model is established and formed in external equipment according to the appearance characteristics of the target and the infrared image of the target;
target detection: after the infrared telescope obtains the infrared image, the infrared telescope enters the training detection model to obtain the target category in the infrared image and a rectangular frame covering the whole target;
target ranging: calculating the distance between the target and an observer according to the preset height of the target to be detected, the pixel height of the target person or object in the picture, the known lens focal length of the infrared telescope and the pixel size of the detector;
image identification: obtaining a current target distance to form a target distance identifier;
and (4) observation and display: the image on the electronic display screen is projected onto the retina of the human eye through the ocular lens formed by arranging the infrared visible light glass lenses with different appearances and different refractive indexes.
The further technical scheme of the invention is as follows: according to the preset height h of the target to be measured0The pixel height of the target person or object in the picture is q, the lens focal length f of the known infrared telescope and the pixel size d of the detector are known, and the distance L between the current target person or object and an observer can be calculated by the following formula:
the further technical scheme of the invention is as follows: and forming a category identifier for the acquired category of the current target, and superposing the target category identifier and the distance identifier on the infrared image in a screen menu manner.
The technical scheme of the invention is as follows: provides an infrared telescope distance measuring method based on external equipment training,
the method for completing the establishment of the training detection model in the external equipment comprises the following steps:
determining the category of the target to be detected: classifying the target to be detected according to the appearance characteristics of the target;
acquiring infrared image data: the method comprises the steps of obtaining an infrared image of a target to be detected through shooting of infrared image shooting equipment, and enabling target images of various categories to form infrared image data collection;
calibrating a data set: after the infrared data set is obtained, completing calibration of the target to be detected in the infrared image, wherein the calibration content comprises the category of each target to be detected and a rectangular frame capable of covering the whole target;
training a detection model: establishing a detection model, and taking the calibrated data as input data to obtain complete model parameters;
the method for measuring the distance in the infrared telescope comprises the following steps:
model import: aiming at the used infrared telescope embedded hardware platform, after converting the training detection model into a format matched with the embedded platform, the embedded system of the infrared telescope is led into the training detection model;
target detection: after the infrared telescope obtains the infrared image, the infrared telescope enters the training detection model to obtain the target category in the infrared image and a rectangular frame covering the whole target;
target ranging: calculating the distance between the target and an observer according to the preset height of the target to be detected, the pixel height of the target person or object in the picture, the known lens focal length of the infrared telescope and the pixel size of the detector;
image identification: obtaining a current target distance to form a target distance identifier;
and (4) observation and display: the image on the electronic display screen is projected onto the retina of the human eye through the ocular lens formed by arranging the infrared visible light glass lenses with different appearances and different refractive indexes.
The further technical scheme of the invention is as follows: the infrared image in the infrared image data acquisition is divided into a sample infrared image and a background infrared image.
The further technical scheme of the invention is as follows: and after the data set is calibrated, the infrared data set is augmented, a new infrared data sample is generated through the original infrared data sample, and the augmentation of the infrared data sample comprises single-image processing and multi-image processing.
The technical scheme of the invention is as follows: the infrared telescope range finding system comprises an infrared imaging lens, an infrared detector, an embedded processor module, an electronic display screen and an eyepiece module, wherein infrared glass lenses with different curvatures and different refractive indexes are arranged on the same axis in an infrared telescope to form the infrared imaging lens, an infrared photosensitive unit is arranged into an uncooled infrared focal plane array to form the infrared detector in the infrared telescope, the embedded processor module comprises a target detection module, a target range finding module and an image identification module, and aiming at an embedded hardware platform of the infrared telescope, a training detection model is converted into a format matched with the embedded platform, and then the embedded system of the infrared telescope is led into the training detection model; the training detection model is established and formed on external equipment according to the appearance characteristics of a target and an infrared image of the target, the infrared telescope enters the training detection model after acquiring the infrared image, the target detection module acquires the target category in the infrared image and a rectangular frame covering the whole target, the target ranging module calculates the distance between the target and an observer according to the height of a preset target to be detected, the pixel height of a target person or object in a picture, the lens focal length of a known infrared telescope and the pixel size of a detector, the image identification module acquires the current target distance to form a target distance identification, infrared visible light glass lenses with different appearances and different refractive indexes are arranged to form an eyepiece module, and the eyepiece module projects the image on an electronic display screen onto the retina of human eyes.
The further technical scheme of the invention is as follows: the eyepiece module changes the position of one or more visible light glass lenses through the lens position adjusting mechanism, thereby changing the virtual image position of the electronic display screen.
The further technical scheme of the invention is as follows: the system also comprises an interface management unit, wherein the interface management unit manages the receiving of the infrared detector signals, the receiving of the I/O signals of the key module and the data output to the electronic display screen.
The further technical scheme of the invention is as follows: the system also comprises a non-uniformity correction module, wherein the non-uniformity correction module adopts an embedded processor module to carry out non-uniformity correction on the array digital signals of the infrared detector so as to correct the temperature response rate of the photosensitive unit of the infrared detector.
The technical scheme of the invention is as follows: the infrared telescope ranging system based on the external equipment is constructed and is characterized by comprising the external equipment and an infrared telescope, wherein the external equipment comprises a training detection model generation module, the training detection model generation module classifies targets to be detected according to the appearance characteristics of the targets in the process of generating a detection model, infrared images of the targets to be detected are obtained through shooting of infrared image shooting equipment, the target images of various types form infrared image data collection, after an infrared data set is obtained, the calibration of the targets to be detected in the infrared images is completed, calibration contents comprise the type of each target to be detected and a rectangular frame capable of covering the whole target, a detection model is established, and the calibrated data are used as input data to obtain complete model parameters; the infrared telescope includes: the infrared imaging lens, the infrared detector, the embedded processor module, the electronic display screen, the eyepiece module and the infrared glass lenses with different curvatures and different refractive indexes are arranged on the same axis in the infrared telescope to form the infrared imaging lens, the infrared photosensitive unit is set into an uncooled infrared focal plane array to form the infrared detector in the infrared telescope, the embedded processor module comprises a target detection module, a target distance measurement module and an image identification module, aiming at the used embedded hardware platform of the infrared telescope, a training detection model is converted into a format matched with the embedded platform, the embedded system of the infrared telescope is led into the training detection model, the infrared telescope enters the training detection model after obtaining an infrared image, the target detection module obtains the target type in the infrared image and a rectangular frame covering the whole target, the target distance measuring module calculates the distance between a target and an observer according to the preset height of the target to be measured, the pixel height of a target person or object in a picture, the known lens focal length of the infrared telescope and the detector pixel size, the image identification module obtains the current target distance to form a target distance identification, the image identification module obtains an eyepiece module formed by arranging infrared visible light glass lenses with different shapes and different refractive indexes, and the eyepiece module projects an image on the electronic display screen onto the retina of a human eye.
The invention has the technical effects that: the method and the system for measuring the distance of the infrared telescope based on the training of the external equipment are constructed, infrared glass lenses with different curvatures and different refractive indexes are arranged on the same axis in the infrared telescope to form an infrared imaging lens, an infrared photosensitive unit is arranged into an uncooled infrared focal plane array to form an infrared detector in the infrared telescope, and the establishment of a training detection model is completed in the external equipment, and the method comprises the following steps: determining the category of the target to be detected: classifying the target to be detected according to the appearance characteristics of the target; acquiring infrared image data: the method comprises the steps of obtaining an infrared image of a target to be detected through shooting of infrared image shooting equipment, and enabling target images of various categories to form infrared image data collection; calibrating a data set: after the infrared data set is obtained, completing calibration of the target to be detected in the infrared image, wherein the calibration content comprises the category of each target to be detected and a rectangular frame capable of covering the whole target; training a detection model: establishing a detection model, and taking the calibrated data as input data to obtain complete model parameters; the method for measuring the distance in the infrared telescope comprises the following steps: model import: aiming at the used infrared telescope embedded hardware platform, after converting the training detection model into a format matched with the embedded platform, the embedded system of the infrared telescope is led into the training detection model; target detection: after the infrared telescope obtains the infrared image, the infrared telescope enters the training detection model to obtain the target category in the infrared image and a rectangular frame covering the whole target; target ranging: calculating the distance between the target and an observer according to the preset height of the target to be detected, the pixel height of the target person or object in the picture, the known lens focal length of the infrared telescope and the pixel size of the detector; image identification: obtaining a current target distance to form a target distance identifier; and (4) observation and display: the image on the electronic display screen is projected onto the retina of the human eye through the ocular lens formed by arranging the infrared visible light glass lenses with different appearances and different refractive indexes. The method comprises the steps of acquiring an infrared data set, calibrating the data set and training a model to obtain a detection model for target detection, and performing category identification, target ranging and image identification on an infrared target by using the detection model and then performing identification superposition to complete real-time ranging and labeling of the infrared image target.
Drawings
FIG. 1 is a schematic view of an infrared telescope according to the present invention.
FIG. 2 is a schematic diagram of the infrared telescope and the training detection model generation module according to the present invention.
FIG. 3 is a block diagram of an embedded processor module according to the present invention.
Detailed Description
The technical solution of the present invention is further illustrated below with reference to specific examples.
The technical implementation mode of the invention is as follows: the method for measuring the distance of the infrared telescope is constructed, infrared glass lenses with different curvatures and different refractive indexes are arranged on the same axis in the infrared telescope to form an infrared imaging lens, an infrared photosensitive unit is arranged into an uncooled infrared focal plane array to form an infrared detector in the infrared telescope, and the method comprises the following steps:
model import: aiming at the used infrared telescope embedded hardware platform, after converting the training detection model into a format matched with the embedded platform, the embedded system of the infrared telescope is led into the training detection model; and the training detection model is established and formed in external equipment according to the appearance characteristics of the target and the infrared image of the target.
The specific implementation mode is as follows: an image processing chip is used as an embedded hardware platform, a detection model format matched with the embedded hardware platform is a wk format, a standard deep learning frame is needed for generating wk format files, a frame supported by the image processing chip is a Caffe frame (configurable Architecture for Fast Feature Embedding Convolutional neural network frame), in a specific embodiment, a Yolov3 model needs to be converted into the Caffe frame through a model conversion tool such as a RuyiStudio application program, and corresponding wk format files are generated under the Caffe frame; and importing the wk file into a memory of an image processing chip, and calling the wk file by software in the process of realizing target detection.
Target detection: and after the infrared telescope obtains the infrared image, the infrared telescope enters the training detection model to obtain the target category in the infrared image and a rectangular frame covering the whole target.
The specific implementation mode is as follows: the infrared telescope is aligned to a scene including a target to be identified, after a target detection function is started by software, a wk model file is called from a memory of the embedded chip to generate a detection model, and the software loads an obtained infrared image into the detection model to calculate so as to obtain a target category in the infrared image and a rectangular frame covering the whole target.
Target ranging: and calculating the distance between the target and an observer according to the preset height of the target to be detected, the pixel height of the target person or object in the picture, the known lens focal length of the infrared telescope and the pixel size of the detector.
The specific implementation mode is as follows: the pixel height of a rectangular frame covering the whole target obtained by target detection is regarded as the pixel height of a target person or object in a picture as q, and the height h of the target to be detected is preset0When the lens focal length f and the detector pixel size d of the infrared telescope are known, the distance L between the current target person or object and the observer can be calculated by the following formula:
taking a detected target person as an example, the height of a pixel of a rectangular frame covering the target person obtained by target detection is 100 pixels, the height of the preset person is 1.7m, the focal length of an infrared telescope lens is 20mm, the pixel size of a detector is 17um, and according to a formula, the distance between the target person and an observer is as follows:
image identification: and forming a category identifier according to the acquired category of the current target, obtaining a distance between the current target and a target distance identifier, and overlapping the target category identifier and the distance identifier on the infrared image in a screen menu mode.
The specific implementation mode is as follows: the category of the target can be but is not limited to English or Chinese display, the target distance can be but is not limited to meter as a distance unit, the target category and the target distance can be displayed near the target, such as the position right above the target, or can be displayed around the screen, such as the left side of the screen, and the target and the identification are related through line segments; the logo display area may use a transparent background, or may use a gray scale background or other colored background.
And (4) observation and display: the image on the electronic display screen is projected onto the retina of the human eye through the ocular lens formed by arranging the infrared visible light glass lenses with different appearances and different refractive indexes.
The technical implementation mode of the invention is as follows: an infrared telescope ranging method based on external equipment training is provided.
The training detection model is established in an external device, wherein the external device can be an external processing system device with strong processing capacity, such as a PC (personal computer), a notebook computer and the like. The method comprises the following steps:
determining the category of the target to be detected: and classifying the target to be detected according to the appearance characteristics of the target.
The specific implementation process is as follows: the method comprises the steps of classifying objects to be detected according to object appearance characteristics, wherein the object types to be detected refer to object types with the same physical characteristics, such as people, cats, chairs and the like, and the number of the object types to be detected is a finite value.
Acquiring infrared image data: the infrared image of the target to be detected is obtained through shooting by the infrared image shooting equipment, and the target images of various categories form infrared image data collection.
The specific implementation process is as follows: the infrared image in the infrared image data acquisition is divided into a sample infrared image and a background infrared image, the sample infrared image should contain one or more objects in the category of the target to be detected, and the background infrared image does not contain the category of the target to be detected. In the infrared image data acquisition process, for the selection of color modes such as black heat, white heat, pseudo color and the like, one or more color modes are selected according to the color mode determination used in the actual detection process.
Calibrating a data set: and after the infrared data set is obtained, completing the calibration of the target to be detected in the infrared image, wherein the calibration content comprises the category of each target to be detected and a rectangular frame capable of covering the whole target.
The specific implementation mode is as follows: after the infrared data set is obtained, the calibration of the target to be detected in the infrared image is completed through operation, and the calibration content comprises the category of each target to be detected and a rectangular frame capable of covering the whole target. After the data set is calibrated, the data set is augmented, a new data sample is generated through an original infrared data sample, and the sample augmentation comprises single-image processing and multi-image processing; the single image processing is to form a new infrared image by adjusting the brightness, contrast, hue, saturation and noise of the original infrared image, randomly increasing scale change, cutting, turning, rotating and the like, and adjust a response target information text file as a new infrared image sample; the multi-image processing is to perform weight addition on pixel values of two infrared images or to perform cutting and splicing on a plurality of images to the same infrared image to form a new infrared image, and adjust a responded target information text file to be used as a new infrared image sample.
In a specific embodiment, the operation may be a manual calibration operation, and the formed rectangular frame is a minimum rectangular frame as much as possible, where the "minimum rectangular frame" is not infinitely close to the meaning of the "minimum" that cannot be determined, but is a "minimum rectangular frame" that can be operated in a specific operation process. The specific method comprises the following steps: the upper boundary position of the minimum rectangular frame is the uppermost pixel of the target to be detected, the lower boundary position is the lowermost pixel of the target to be detected, the left boundary position is the leftmost pixel of the target to be detected, and the right boundary position is the rightmost pixel of the target to be detected; after calibration, each infrared image corresponds to a text file, all information of the target to be detected in the infrared image is stored in the file, the information of the target to be detected comprises the category of the target and the position coordinate of the minimum rectangular frame covering the target in the infrared image, the position coordinate usually takes the upper left corner of the infrared image as the origin of coordinates, the right corner of the infrared image as the positive direction of horizontal coordinates and the downward corner of the infrared image as the positive direction of vertical coordinates, and the position coordinate of the rectangular frame can be determined by using coordinates of the upper left corner and the lower right corner and can also be determined by using the coordinates of the upper left corner and the long and wide pixel value of the rectangular frame.
Training a detection model: and establishing a detection model, and taking the calibrated data as input data to obtain complete model parameters.
The specific implementation mode is as follows: the training detection model is based on a Yolov3 feature extraction network, and three deep convolutional layers, four maximum pooling layers with different scales and a residual error network module consisting of 1 and 5 deep convolutional layer branches are added behind a backbone network Darknet-53.
The method for measuring the distance in the infrared telescope comprises the following steps:
model import: and aiming at the used embedded hardware platform, the model is converted into a format matched with the embedded platform and then is imported into an embedded system.
The specific implementation mode is as follows: aiming at the used infrared telescope embedded hardware platform, after converting the training detection model into a format matched with the embedded platform, the embedded system of the infrared telescope is led into the training detection model; and the training detection model is established and formed in external equipment according to the appearance characteristics of the target and the infrared image of the target. An image processing chip is used as an embedded hardware platform, a detection model format matched with the embedded hardware platform is a wk format, a standard deep learning frame is needed for generating wk format files, a frame supported by the image processing chip is a Caffe frame (configurable Architecture for Fast Feature Embedding Convolutional neural network frame), in a specific embodiment, a Yolov3 model needs to be converted into the Caffe frame through a model conversion tool such as a RuyiStudio application program, and corresponding wk format files are generated under the Caffe frame; and importing the wk file into a memory of an image processing chip, and calling the wk file by software in the process of realizing target detection.
Target detection: and after the infrared telescope obtains the infrared image, the infrared telescope enters the training detection model to obtain the target category in the infrared image and a rectangular frame covering the whole target.
The specific implementation mode is as follows: the infrared telescope is aligned to a scene including a target to be identified, after a target detection function is started by software, a wk model file is called from a memory of the embedded chip to generate a detection model, and the software loads an obtained infrared image into the detection model to calculate so as to obtain a target category in the infrared image and a rectangular frame covering the whole target.
Target ranging: and calculating the distance between the target and an observer according to the preset height of the target to be detected, the pixel height of the target person or object in the picture, the known lens focal length of the infrared telescope and the pixel size of the detector.
The specific implementation mode is as follows: the pixel height of a rectangular frame covering the whole target obtained by target detection is regarded as the pixel height of a target person or object in a picture as q, and the height h of the target to be detected is preset0When the lens focal length f and the detector pixel size d of the infrared telescope are known, the distance L between the current target person or object and the observer can be calculated by the following formula:
taking a detected target person as an example, the height of a pixel of a rectangular frame covering the target person obtained by target detection is 100 pixels, the height of the preset person is 1.7m, the focal length of an infrared telescope lens is 20mm, the pixel size of a detector is 17um, and according to a formula, the distance between the target person and an observer is as follows:
image identification: and forming a category identifier according to the acquired category of the current target, obtaining a distance between the current target and a target distance identifier, and overlapping the target category identifier and the distance identifier on the infrared image in a screen menu mode.
The specific implementation mode is as follows: the category of the target can be but is not limited to English or Chinese display, the target distance can be but is not limited to meter as a distance unit, the target category and the target distance can be displayed near the target, such as the position right above the target, or can be displayed around the screen, such as the left side of the screen, and the target and the identification are related through line segments; the logo display area may use a transparent background, or may use a gray scale background or other colored background.
And (4) observation and display: the image on the electronic display screen is projected onto the retina of the human eye through the ocular lens formed by arranging the infrared visible light glass lenses with different appearances and different refractive indexes.
As shown in fig. 2 and 3, the embodiment of the present invention is: an infrared telescope ranging system is constructed and comprises an infrared imaging lens 1, an infrared detector 2, an embedded processor module 3, an electronic display screen 4 and an eyepiece module 5, infrared glass lenses with different curvatures and different refractive indexes are arranged on the same axis in an infrared telescope to form the infrared imaging lens 1, an infrared photosensitive unit is arranged into an uncooled infrared focal plane array to form the infrared detector 2 in the infrared telescope, the embedded processor module 3 comprises a target detection module 31, a target ranging module 32 and an image identification module 33, aiming at an infrared telescope embedded hardware platform, a training detection model is converted into a format matched with the embedded platform, the embedded system of the infrared telescope is led into the training detection model, the infrared telescope enters the training detection model after acquiring an infrared image, the target detection module 31 acquires a target type in the infrared image and a rectangular frame covering the whole target, the target ranging module 32 calculates the distance between a target and an observer according to the preset height of the target to be measured, the pixel height of a target person or object in a picture, the lens focal length of a known infrared telescope and the pixel size of a detector, the image identification module 33 obtains the current target distance to form a target distance identification, the eyepiece module 5 is formed by arranging infrared visible light glass lenses with different shapes and different refractive indexes, and the eyepiece module 5 projects an image on the electronic display screen onto the retina of a human eye.
The specific implementation process of the invention is as follows: an image processing chip is used as an embedded hardware platform, a detection model format matched with the embedded hardware platform is a wk format, a standard deep learning frame is needed for generating wk format files, a frame supported by the image processing chip is a Caffe frame (configurable Architecture for Fast Feature Embedding Convolutional neural network frame), in a specific embodiment, a Yolov3 model needs to be converted into the Caffe frame through a model conversion tool such as a RuyiStudio application program, and corresponding wk format files are generated under the Caffe frame; and importing the wk file into a memory of an image processing chip, and calling the wk file by software in the process of realizing target detection. The infrared telescope is aligned to a scene including a target to be identified, after a target detection function is started by software, a wk model file is called from a memory of an embedded chip to generate a detection model, the obtained infrared image is loaded into the detection model by the software for calculation, and the target detection module 31 obtains the target category in the infrared image and a rectangular frame covering the whole target. The pixel height of a rectangular frame covering the whole target obtained by target detection is regarded as the pixel height of a target person or object in a picture as q, and the height h of the target to be detected is preset0When the lens focal length f and the detector pixel size d of the infrared telescope are known, the distance L between the current target person or object and the observer can be calculated by the following formula:
taking a detected target person as an example, the height of a pixel of a rectangular frame covering the target person obtained by target detection is 100 pixels, the height of the preset person is 1.7m, the focal length of an infrared telescope lens is 20mm, the pixel size of a detector is 17um, and according to a formula, the distance between the target person and an observer is as follows:
the target ranging module 32 calculates the distance between the target and the observer according to the preset height of the target to be measured, the pixel height of the target person or object in the picture, the known lens focal length of the infrared telescope and the pixel size of the detector.
The category of the target can be but is not limited to English or Chinese display, the target distance can be but is not limited to meter as a distance unit, the target category and the target distance can be displayed near the target, such as the position right above the target, or can be displayed around the screen, such as the left side of the screen, and the target and the identification are related through line segments; the logo display area may use a transparent background, or may use a gray scale background or other colored background. The image identification module 33 obtains the current target distance to form a target distance identification, the eyepiece module 5 is formed by arranging infrared visible light glass lenses with different shapes and different refractive indexes, and the eyepiece module 5 projects an image on the electronic display screen onto the retina of a human eye.
The training detection model generation module 6 classifies the targets to be detected according to the appearance characteristics of the targets in the detection model generation process, obtains infrared images of the targets to be detected by shooting through infrared image shooting equipment, forms the target images of various categories into infrared image data to be collected, finishes the calibration of the targets to be detected in the infrared images after obtaining an infrared data set, establishes a detection model by taking the calibrated data as input data, and obtains complete model parameters;
the preferred embodiments of the present invention are: the eyepiece module 5 changes the position of one or more visible light glass lenses through the lens position adjusting mechanism, so that the virtual image position of the electronic display screen is changed. The eyepiece module may use a mirror.
The preferred embodiments of the present invention are: the system also comprises an interface management unit, wherein the interface management unit manages the receiving of the infrared detector signals, the receiving of the I/O signals of the key module and the data output to the electronic display screen. The system also comprises a non-uniformity correction module, wherein the non-uniformity correction module adopts an embedded processor module to carry out non-uniformity correction on the array digital signals of the infrared detector so as to correct the temperature response rate of the photosensitive unit of the infrared detector.
The technical scheme of the invention is as follows: the infrared telescope ranging system based on the external equipment is constructed and is characterized by comprising the external equipment and an infrared telescope, wherein the external equipment comprises a training detection model generation module 6, the training detection model generation module 6 classifies targets to be detected according to the appearance characteristics of the targets in the process of generating a detection model, infrared images of the targets to be detected are obtained through the shooting of infrared image shooting equipment, the target images of various categories form infrared image data acquisition, after an infrared data set is obtained, the calibration of the targets to be detected in the infrared images is completed, calibration contents comprise the category of each target to be detected and a rectangular frame capable of covering the whole target, a detection model is established, and the calibrated data are used as input data to obtain complete model parameters; the infrared telescope includes: an infrared imaging lens 1, an infrared detector 2, an embedded processor module 3, an electronic display screen 4, an ocular module 5, infrared glass lenses with different curvatures and different refractive indexes are arranged on the same axis in an infrared telescope to form the infrared imaging lens 1, an infrared photosensitive unit is set into a non-refrigeration infrared focal plane array to form the infrared detector 2 in the infrared telescope, the embedded processor module 3 comprises a target detection module 31, a target ranging module 32 and an image identification module 33, aiming at an embedded hardware platform of the infrared telescope, a training detection model is converted into a format matched with the embedded platform, an embedded system of the infrared telescope is led into the training detection model, the infrared telescope enters the training detection model after obtaining an infrared image, the target detection module 31 obtains the target category in the infrared image and a rectangular frame covering the whole target, the target ranging module 32 calculates the distance between a target and an observer according to the preset height of the target to be measured, the pixel height of a target person or object in a picture, the lens focal length of a known infrared telescope and the pixel size of a detector, the image identification module 33 obtains the current target distance to form a target distance identification, the eyepiece module 5 is formed by arranging infrared visible light glass lenses with different shapes and different refractive indexes, and the eyepiece module 5 projects an image on the electronic display screen onto the retina of a human eye.
The method for completing the establishment of the training detection model in the external equipment comprises the following steps:
the method comprises the steps of classifying objects to be detected according to object appearance characteristics, wherein the object types to be detected refer to object types with the same physical characteristics, such as people, cats, chairs and the like, and the number of the object types to be detected is a finite value. The infrared image in the infrared image data acquisition is divided into a sample infrared image and a background infrared image, the sample infrared image should contain one or more objects in the category of the target to be detected, and the background infrared image does not contain the category of the target to be detected. In the infrared image data acquisition process, for the selection of color modes such as black heat, white heat, pseudo color and the like, one or more color modes are selected according to the color mode determination used in the actual detection process. After the infrared data set is obtained, the calibration of the target to be detected in the infrared image is completed through operation, and the calibration content comprises the category of each target to be detected and a rectangular frame capable of covering the whole target. After the data set is calibrated, the data set is augmented, a new data sample is generated through an original infrared data sample, and the sample augmentation comprises single-image processing and multi-image processing; the single image processing is to form a new infrared image by adjusting the brightness, contrast, hue, saturation and noise of the original infrared image, randomly increasing scale change, cutting, turning, rotating and the like, and adjust a response target information text file as a new infrared image sample; the multi-image processing is to perform weight addition on pixel values of two infrared images or to perform cutting and splicing on a plurality of images to the same infrared image to form a new infrared image, and adjust a responded target information text file to be used as a new infrared image sample.
In a specific embodiment, the operation may be a manual calibration operation, and the formed rectangular frame is a minimum rectangular frame as much as possible, where the "minimum rectangular frame" is not infinitely close to the meaning of the "minimum" that cannot be determined, but is a "minimum rectangular frame" that can be operated in a specific operation process. The specific method comprises the following steps: the upper boundary position of the minimum rectangular frame is the uppermost pixel of the target to be detected, the lower boundary position is the lowermost pixel of the target to be detected, the left boundary position is the leftmost pixel of the target to be detected, and the right boundary position is the rightmost pixel of the target to be detected; after calibration, each infrared image corresponds to a text file, all information of the target to be detected in the infrared image is stored in the file, the information of the target to be detected comprises the category of the target and the position coordinate of the minimum rectangular frame covering the target in the infrared image, the position coordinate usually takes the upper left corner of the infrared image as the origin of coordinates, the right corner of the infrared image as the positive direction of horizontal coordinates and the downward corner of the infrared image as the positive direction of vertical coordinates, and the position coordinate of the rectangular frame can be determined by using coordinates of the upper left corner and the lower right corner and can also be determined by using the coordinates of the upper left corner and the long and wide pixel value of the rectangular frame. And establishing a detection model, and taking the calibrated data as input data to obtain complete model parameters. The specific implementation mode is as follows: the training detection model is based on a Yolov3 feature extraction network, and three deep convolutional layers, four maximum pooling layers with different scales and a residual error network module consisting of 1 and 5 deep convolutional layer branches are added behind a backbone network Darknet-53.
The invention has the technical effects that: the method for measuring the distance of the infrared telescope comprises the following steps: determining the category of the target to be detected: classifying the target to be detected according to the appearance characteristics of the target; acquiring infrared image data: the method comprises the steps of obtaining an infrared image of a target to be detected through shooting of infrared image shooting equipment, and enabling target images of various categories to form infrared image data collection; calibrating a data set: after the infrared data set is obtained, completing calibration of the target to be detected in the infrared image, wherein the calibration content comprises the category of each target to be detected and a rectangular frame capable of covering the whole target; training a detection model: establishing a detection model, and taking the calibrated data as input data to obtain complete model parameters; model import: aiming at the used embedded hardware platform, the model is converted into a format matched with the embedded platform and then is led into an embedded system; target detection: after the infrared telescope obtains the infrared image, the infrared telescope enters the training detection model to obtain the target category in the infrared image and a rectangular frame covering the whole target; target ranging: calculating the distance between the target and an observer according to the preset height of the target to be detected, the pixel height of the target person or object in the picture, the known lens focal length of the infrared telescope and the pixel size of the detector; image identification: obtaining a current target distance to form a target distance identifier; and (4) observation and display: the image on the electronic display screen is projected onto the retina of the human eye through the ocular lens formed by arranging the infrared visible light glass lenses with different appearances and different refractive indexes. The method comprises the steps of acquiring an infrared data set, calibrating the data set and training a model to obtain a detection model for target detection, and performing category identification, target ranging and image identification on an infrared target by using the detection model and then performing identification superposition to complete real-time ranging and labeling of the infrared image target.
The foregoing is a more detailed description of the invention in connection with specific preferred embodiments and it is not intended that the invention be limited to these specific details. For those skilled in the art to which the invention pertains, several simple deductions or substitutions can be made without departing from the spirit of the invention, and all shall be considered as belonging to the protection scope of the invention.
Claims (10)
1. The distance measuring method of the infrared telescope is characterized by comprising the following steps of:
model import: aiming at the used infrared telescope embedded hardware platform, after converting the training detection model into a format matched with the embedded platform, the embedded system of the infrared telescope is led into the training detection model; the training detection model is established and formed in external equipment according to the appearance characteristics of the target and the infrared image of the target;
target detection: after the infrared telescope obtains the infrared image, the infrared telescope enters the training detection model to obtain the target category in the infrared image and a rectangular frame covering the whole target;
target ranging: calculating the distance between the target and an observer according to the preset height of the target to be detected, the pixel height of the target person or object in the picture, the known lens focal length of the infrared telescope and the pixel size of the detector;
image identification: obtaining a current target distance to form a target distance identifier;
and (4) observation and display: the image on the electronic display screen is projected onto the retina of the human eye through the ocular lens formed by arranging the infrared visible light glass lenses with different appearances and different refractive indexes.
2. The infrared telescope distance measuring method according to claim 1, wherein the height h of the target to be measured is preset0The pixel height of the target person or object in the picture is q, the lens focal length f of the known infrared telescope and the pixel size d of the detector are known, and the distance L between the current target person or object and an observer can be calculated by the following formula:
3. the infrared telescope ranging method according to claim 1, wherein the obtained category of the current target is formed into a category identifier, and the target category identifier and the distance identifier are superimposed on the infrared image in an on-screen menu manner.
4. An infrared telescope distance measuring method based on external equipment training is characterized in that,
the method for completing the establishment of the training detection model in the external equipment comprises the following steps:
determining the category of the target to be detected: classifying the target to be detected according to the appearance characteristics of the target;
acquiring infrared image data: the method comprises the steps of obtaining an infrared image of a target to be detected through shooting of infrared image shooting equipment, and enabling target images of various categories to form infrared image data collection;
calibrating a data set: after the infrared data set is obtained, completing calibration of the target to be detected in the infrared image, wherein the calibration content comprises the category of each target to be detected and a rectangular frame capable of covering the whole target;
training a detection model: establishing a detection model, and taking the calibrated data as input data to obtain complete model parameters;
the method for measuring the distance in the infrared telescope comprises the following steps:
model import: aiming at the used infrared telescope embedded hardware platform, after converting the training detection model into a format matched with the embedded platform, the embedded system of the infrared telescope is led into the training detection model; (ii) a
Target detection: after the infrared telescope obtains the infrared image, the infrared telescope enters the training detection model to obtain the target category in the infrared image and a rectangular frame covering the whole target;
target ranging: calculating the distance between the target and an observer according to the preset height of the target to be detected, the pixel height of the target person or object in the picture, the known lens focal length of the infrared telescope and the pixel size of the detector;
image identification: obtaining a current target distance to form a target distance identifier;
and (4) observation and display: the image on the electronic display screen is projected onto the retina of the human eye through the ocular lens formed by arranging the infrared visible light glass lenses with different appearances and different refractive indexes.
5. The infrared telescope ranging method based on external device training as claimed in claim 4, wherein the infrared image in the infrared image data collection is divided into a sample infrared image and a background infrared image.
6. The infrared telescope ranging method based on external device training as claimed in claim 4, wherein after calibration of the data set, the infrared data set is augmented, new infrared data samples are generated from original infrared data samples, and the augmentation of the infrared data samples includes single-image processing and multi-image processing.
7. An infrared telescope ranging system is characterized by comprising an infrared imaging lens, an infrared detector, an embedded processor module, an electronic display screen and an eyepiece module, wherein infrared glass lenses with different curvatures and different refractive indexes are arranged on the same axis in an infrared telescope to form the infrared imaging lens, an infrared photosensitive unit is arranged into an uncooled infrared focal plane array to form the infrared detector in the infrared telescope, the embedded processor module comprises a target detection module, a target ranging module and an image identification module, and aiming at an infrared telescope embedded hardware platform, a training detection model is converted into a format matched with the embedded platform, and then the embedded system of the infrared telescope is led into the training detection model; the training detection model is established and formed on external equipment according to the appearance characteristics of a target and an infrared image of the target, the infrared telescope enters the training detection model after acquiring the infrared image, the target detection module acquires the target category in the infrared image and a rectangular frame covering the whole target, the target ranging module calculates the distance between the target and an observer according to the height of a preset target to be detected, the pixel height of a target person or object in a picture, the lens focal length of a known infrared telescope and the pixel size of a detector, the image identification module acquires the current target distance to form a target distance identification, infrared visible light glass lenses with different appearances and different refractive indexes are arranged to form an eyepiece module, and the eyepiece module projects the image on an electronic display screen onto the retina of human eyes.
8. The infrared telescope ranging system of claim 7, wherein the eyepiece module changes the position of one or more of the visible light glass lenses via a lens position adjustment mechanism, thereby changing the position of the virtual image of the electronic display.
9. The infrared telescope ranging system of claim 7, further comprising an interface management unit that manages the reception of infrared detector signals, the reception of key module I/O signals, and the output of data to an electronic display screen.
10. An infrared telescope ranging system based on external equipment is characterized by comprising the external equipment and an infrared telescope, wherein the external equipment comprises a training detection model generation module, the training detection model generation module classifies targets to be detected according to target appearance characteristics in the process of generating a detection model, infrared images of the targets to be detected are obtained through shooting of infrared image shooting equipment, the target images of various types form infrared image data collection, after an infrared data set is obtained, calibration of the targets to be detected in the infrared images is completed, calibration contents comprise the type of each target to be detected and a rectangular frame capable of covering the whole target, a detection model is established, and the calibrated data are used as input data to obtain complete model parameters; the infrared telescope includes: the infrared imaging lens, the infrared detector, the embedded processor module, the electronic display screen, the eyepiece module and the infrared glass lenses with different curvatures and different refractive indexes are arranged on the same axis in the infrared telescope to form the infrared imaging lens, the infrared photosensitive unit is set into an uncooled infrared focal plane array to form the infrared detector in the infrared telescope, the embedded processor module comprises a target detection module, a target distance measurement module and an image identification module, aiming at the used embedded hardware platform of the infrared telescope, a training detection model is converted into a format matched with the embedded platform, the embedded system of the infrared telescope is led into the training detection model, the infrared telescope enters the training detection model after obtaining an infrared image, the target detection module obtains the target type in the infrared image and a rectangular frame covering the whole target, the target distance measuring module calculates the distance between a target and an observer according to the preset height of the target to be measured, the pixel height of a target person or object in a picture, the known lens focal length of the infrared telescope and the detector pixel size, the image identification module obtains the current target distance to form a target distance identification, the image identification module obtains an eyepiece module formed by arranging infrared visible light glass lenses with different shapes and different refractive indexes, and the eyepiece module projects an image on the electronic display screen onto the retina of a human eye.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111554027.1A CN114295108A (en) | 2021-12-17 | 2021-12-17 | Distance measurement method and system for external equipment and infrared telescope |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111554027.1A CN114295108A (en) | 2021-12-17 | 2021-12-17 | Distance measurement method and system for external equipment and infrared telescope |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114295108A true CN114295108A (en) | 2022-04-08 |
Family
ID=80966839
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111554027.1A Pending CN114295108A (en) | 2021-12-17 | 2021-12-17 | Distance measurement method and system for external equipment and infrared telescope |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114295108A (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2008060319A2 (en) * | 2006-03-27 | 2008-05-22 | Horus Vision Llc | Apparatus and method for calculating aiming point information |
| CN103926010A (en) * | 2014-04-18 | 2014-07-16 | 山东神戎电子股份有限公司 | Multifunctional dual-spectrum portable observation instrument |
| CN105203073A (en) * | 2015-09-08 | 2015-12-30 | 上海巨哥电子科技有限公司 | Imager with electronic distance measurement reticle |
| CN113610695A (en) * | 2021-05-07 | 2021-11-05 | 浙江兆晟科技股份有限公司 | Infrared telescope full-frame imaging output method and system |
-
2021
- 2021-12-17 CN CN202111554027.1A patent/CN114295108A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2008060319A2 (en) * | 2006-03-27 | 2008-05-22 | Horus Vision Llc | Apparatus and method for calculating aiming point information |
| CN103926010A (en) * | 2014-04-18 | 2014-07-16 | 山东神戎电子股份有限公司 | Multifunctional dual-spectrum portable observation instrument |
| CN105203073A (en) * | 2015-09-08 | 2015-12-30 | 上海巨哥电子科技有限公司 | Imager with electronic distance measurement reticle |
| CN113610695A (en) * | 2021-05-07 | 2021-11-05 | 浙江兆晟科技股份有限公司 | Infrared telescope full-frame imaging output method and system |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Burton et al. | Color and spatial structure in natural scenes | |
| CN109377469B (en) | Processing method, system and storage medium for fusing thermal imaging with visible light image | |
| CN105933617B (en) | A kind of high dynamic range images fusion method for overcoming dynamic problem to influence | |
| CN104254869B (en) | For the method and system for the visible light representations for projecting infrared radiation | |
| AU2011242408B2 (en) | Shape and photometric invariants recovery from polarisation images | |
| US20180184062A1 (en) | Image Processing to Determine Center of Balance in a Digital Image | |
| CN107018323B (en) | Control method, control device and electronic device | |
| CN110675448B (en) | Ground lamplight remote sensing monitoring method, system and storage medium based on civil airliner | |
| CN101207832A (en) | Method for checking digital camera color reduction | |
| CN101882034A (en) | Device and method for discriminating color of touch pen of touch device | |
| CN111323125B (en) | Temperature measurement method and device, computer storage medium and electronic equipment | |
| CN109870240A (en) | Infrared Targets image real-time quantitative processing system with environment self-adaption | |
| CN103390290A (en) | Information processing device and information processing method | |
| CN110456904A (en) | A kind of augmented reality glasses eye movement exchange method and system without calibration | |
| US20080166061A1 (en) | Enhanced Infrared Imaging System | |
| CN109497925A (en) | Eye visual function evaluating apparatus and eye Evaluation of visual function | |
| CN107454335B (en) | Image processing method, image processing device, computer-readable storage medium and mobile terminal | |
| CN113358231A (en) | Infrared temperature measurement method, device and equipment | |
| CN111413279A (en) | Video processing method and device for multispectral detection and multispectral detection terminal | |
| CN109738005A (en) | Detection device, detection system and detection method | |
| CN110572576A (en) | Method and system for shooting visible light and thermal imaging overlay image and electronic equipment | |
| CN114295108A (en) | Distance measurement method and system for external equipment and infrared telescope | |
| CN115100556B (en) | Augmented reality method and device based on image segmentation and fusion and electronic equipment | |
| CN115830318A (en) | Method and device for segmenting overlapped tobacco shred images based on improved Mask-RCNN (Richardson network) | |
| Grewe et al. | Low-resolution infrared temperature analysis for disease situation awareness via machine learning on a mobile platform |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |