CN-115601563-B - State detection system of archive storage device

CN115601563BCN 115601563 BCN115601563 BCN 115601563BCN-115601563-B

Abstract

The invention relates to a state detection system of a file storage device, which comprises one or more image acquisition equipment, an image processing unit and an opening state detection unit, wherein the image acquisition equipment is arranged above the file storage device and is used for acquiring a top view of the file storage device, the image processing unit is used for preprocessing the top view of the file storage device acquired by the image acquisition equipment to obtain an image to be detected, and the opening state detection unit of the file storage device is used for obtaining the opening state of the file storage device based on a trained opening position detection model of the file storage device according to the image to be detected obtained by the image processing unit.

Inventors

LI SHU
WANG DONGHAO
SHANG TAO
GAO XIAOQIONG
Yan Guanbo

Assignees

北京京航计算通讯研究所

Dates

Publication Date: 20260505
Application Date: 20221110

Claims (9)

1. A system for detecting a state of an archive storage device, comprising: One or more image acquisition devices mounted above the archive storage device for acquiring a top view of the archive storage device; The image processing unit is used for preprocessing the top view of the archive storage device acquired by the image acquisition equipment to obtain an image to be detected; The file storage device opening state detection unit is used for obtaining the opening state of the file storage device based on the trained file storage device opening position detection model according to the image to be detected obtained by the image processing unit; the archive storage device opening position detection model is a modified ShuffleNet model, the modified ShuffleNet model comprising: The first convolution layer is used for extracting shallow layer features of input data and outputting a shallow layer feature map, and is a one-dimensional convolution layer; The maximum pooling layer is used for reducing the dimension of the shallow feature map; the ShuffleNet modules are used for extracting deep semantic features from the dimension-reduced shallow feature map in a point-by-point group rolling and channel shuffling mode; And the second convolution layer is used for reducing the channel dimension of the deep semantic features and outputting a detection result.
2. A state detection system for an archive storage device of claim 1 wherein, The image processing unit preprocesses the top view of the archive storage device acquired by the image acquisition equipment to obtain an image to be detected, and the image processing unit comprises: In a top view of the archive storage device, a plurality of images with the width W and the length L are extracted as images to be detected in a mode of equal interval in a direction perpendicular to the moving direction of the archive storage device, wherein the long side of each image to be detected is parallel to the moving direction of the archive storage device.
3. The system of claim 1, wherein the image processing unit further comprises an image stitching unit for stitching images acquired by each image acquisition device to obtain a top view of the complete archive storage device when more than one image acquisition device is available.
4. A system for detecting the status of a file storage device according to claim 3, wherein the image stitching unit is adapted to stitch the images acquired by each image acquisition device to obtain a top view of the complete file storage device by: extracting characteristic points in each image by adopting a SIFT algorithm; Performing feature point matching between two adjacent images by adopting a rapid nearest neighbor algorithm to obtain an initial corresponding relation of feature points between the adjacent images; Deleting the error corresponding relation by adopting a random sampling consistency algorithm, and calculating a homography matrix between adjacent images; and mapping different images onto the same image based on homography matrixes among adjacent images to obtain a spliced image.
5. A state detection system for archive storage devices as claimed in claim 1, wherein the Loss of archive storage device opening position detection model is calculated by the following formula: Loss=Point_Loss+Confidence_Loss Where S represents the length of the feature output by the model, k i represents the edge point position in the ith interval of the model prediction, C i represents the confidence that the edge point exists in the ith interval; Indicating whether there is actually a target point in the ith interval, if there is a value of 1, otherwise, 0, λ obj 、λ noobj and λ point indicate weight coefficients.
6. A state detection system of a archive storage device according to claim 1, wherein the archive storage device opening state detection unit includes an edge point detection unit and an opening position calculation unit; the edge point detection unit is used for inputting each image to be detected into the archive storage device opening position detection model to obtain edge points at the archive storage device opening in each image to be detected; The opening position calculating unit is used for calculating the opening position of the file storage device based on the edge points of the opening of the file storage device in each image to be detected.
7. A state detection system of a file storage device according to claim 6, wherein the opening position calculation unit calculates the opening position of the file storage device by: calculating the bottom coordinates of the cabinet corresponding to the edge points of the opening of the file storage device in each image to be detected according to the coordinate projection transformation; And converting the bottom coordinates of the cabinet body at the opening of the file storage device into actual position data according to the internal parameters, the external parameters and the distortion coefficients of the camera.
8. A system for detecting the status of a file storage device as in claim 7, wherein the bottom coordinates (x * ,y * ) of the cabinet at the opening of the file storage device are calculated according to the following formula: Wherein, (x, y) is the coordinates of the edge points of the opening of the file storage device to be detected, (x * ,y * ) is the coordinates of the bottom of the cabinet body at the opening of the file storage device, (x c ,y c ) is the coordinates of the center point of the image, H is the mounting height of the camera, and H is the height of the cabinet body of the file storage device.
9. The system for detecting the state of a file storage device according to claim 7, wherein the following formula is used to convert the bottom coordinates of the cabinet at the opening of the file storage device into actual position data according to the internal parameters, the external parameters and the distortion coefficients of the camera: according to the formula: Converting the pixel coordinates of the edge points into image coordinates (x, y, z), wherein (u, v) represents the pixel coordinates of the edge points, dx and dy represent the actual physical size of the pixels, and (u 0 ,v 0 ) represents the pixel coordinates of the center point of the image; According to the formula Calculating a camera coordinate system lower coordinate (X c ,Y c ,Z c ) of the edge point, wherein f represents a camera focal length; According to the formula World coordinates (X w ,Y w ,Z w ) of the edge points are calculated, where R represents the rotation matrix and T represents the translation matrix.

Description

State detection system of archive storage device Technical Field The present invention relates to the field of detecting openings of archive storage devices, and in particular, to a status detecting system of an archive storage device. Background The file storage device has the advantages of compact structure, small space occupation, large material storage capacity and the like, and is widely applied to warehouses in various industries as important material storage equipment. But in the operation process of the file storage device, due to the problems of large cabinet area, unfixed opening position, shielding caused by high cabinet body and the like, real-time video monitoring of the opening position of the file storage device is difficult, and full-flow monitoring of behaviors of personnel in a warehouse is difficult to realize. In response to this problem, a driving mechanism using a movable rail has been proposed in recent years, which drives a monitoring camera in a small field of view to a position directly above the opening. The core technology that this scheme can realize is to the real-time detection of archives storage device opening part position, how to acquire the opening position of high accuracy, and control hardware cost simultaneously has important meaning to engineering realization. The existing detection method for the opening state of the file storage device is mainly based on a magnetic induction sensor, an ultrasonic sensor, an infrared sensor or a mechanical sensor, and can detect the opening state with higher precision, but the sensor and wiring are required to be installed on each cabinet of the file storage device, so that the cost is higher and the maintenance is difficult. Meanwhile, the cabinet body of the file storage device can move along the guide rail, and the position is not fixed, so that the difficulty of wiring construction is increased, and the service life of wires is seriously influenced. The following problems are presented in detail: 1. and 2, each sensor also needs to be wired, and as the archive storage device cabinet body moves along the guide rail, the difficulty of wiring construction is high, and meanwhile, the service life of the wire is seriously influenced by frequent stretching and bending. Disclosure of Invention In view of the above analysis, an embodiment of the present invention is directed to a state detection system of an archive storage device, so as to solve the problems of high hardware cost and poor maintainability of the existing state detection system of an archive storage device. In one aspect, an embodiment of the present invention provides a state detection system of an archive storage device, including: One or more image acquisition devices mounted above the archive storage device for acquiring a top view of the archive storage device; The image processing unit is used for preprocessing the top view of the archive storage device acquired by the image acquisition equipment to obtain an image to be detected; And the archive storage device opening state detection unit is used for obtaining the opening state of the archive storage device based on the trained archive storage device opening position detection model according to the image to be detected obtained by the image processing unit. Based on the further improvement of the above technical scheme, the image processing unit performs preprocessing on the top view of the archive storage device collected by the image collecting device to obtain an image to be detected, including: In a top view of the archive storage device, a plurality of images with the width W and the length L are extracted as images to be detected in a mode of equal interval in a direction perpendicular to the moving direction of the archive storage device, wherein the long side of each image to be detected is parallel to the moving direction of the archive storage device. The image processing unit further comprises an image splicing unit, wherein the image splicing unit is used for splicing the images acquired by each image acquisition device to obtain a top view of the complete archive storage device when the number of the image acquisition devices is greater than one. Further, the image stitching unit stitch the images acquired by each image acquisition device to obtain a top view of the complete archive storage device by adopting the following steps: extracting characteristic points in each image by adopting a SIFT algorithm; Performing feature point matching between two adjacent images by adopting a rapid nearest neighbor algorithm to obtain an initial corresponding relation of feature points between the adjacent images; Deleting the error corresponding relation by adopting a random sampling consistency algorithm, and calculating a homography matrix between adjacent images; and mapping different images onto the same image based on homography matrixes among adjacent images to obtain a spliced image. Further, the archive s