Self-adaptive hierarchical image segmentation identification method, device and system

Abstract

The invention discloses a self-adaptive hierarchical image segmentation identification method, device and system. The method comprises the following steps: acquiring an original image of a certain frame in a high-definition video; performing image segmentation on the original image to obtain a plurality of image slices with overlapped pixels; carrying out iterative segmentation and identificationon the sub-image slices; and finally, selecting all identification results to obtain an identification result of the final image. According to the method, the problems of low speed and low small object detection rate of existing high-definition image object recognition are solved, and the effects of high object recognition speed and high detection rate in the high-definition image are achieved.

Description

technical field

[0001] The present invention relates to the technical field of high-definition image recognition, in particular to an adaptive hierarchical image segmentation recognition method, device and system. Background technique

[0002] Pedestrian recognition in densely populated places requires the use of high-definition cameras and object recognition technology, while traditional object recognition requires controlling image pixels in a relatively small size to improve recognition efficiency and adapt to the cache requirements of recognition devices. However, compressing images to a fixed size will result in the loss of a large number of small and medium object information, especially for densely populated areas such as train stations and squares. Using high-definition video acquisition equipment such as 4K and 8K resolution for monitoring, it is necessary to solve real-time and accurate object recognition for high-resolution images in video streams.

[0003] High-...

Examples

Embodiment 1

[0054] figure 1 It is a flow chart of an adaptive hierarchical image segmentation and recognition method provided by the embodiment of the present invention; an adaptive hierarchical image segmentation and recognition method provided by this embodiment will now be described in conjunction with an implementation example. The method includes the following steps:

[0055] Acquiring image step S101, obtaining the original image of a certain frame in the high-definition video;

[0056] Specifically, the video stream of a certain scene is captured by a high-definition camera (the resolution of the captured picture is at least 1920*1080), and the video stream data is transmitted to the memory of the management server through a wired network or a wireless network such as 5G. A frame of original image is sampled from the video stream at fixed intervals such as 0.1s. The size of the original image is 3840*2160. For specific images, see figure 2 .

[0057] Segmentation step S102, per...

Embodiment 2

[0085] Such as Figure 5 As shown, this embodiment provides an adaptive level image segmentation and recognition device, which is a virtual device corresponding to the adaptive level image segmentation and recognition method described in Embodiment 1, and the device is equipped to execute the The corresponding functional modules and beneficial effects of the method. The unit includes:

[0086] Obtaining an image module 901, configured to obtain an original image of a certain frame in the high-definition video;

[0087] Segmentation module 902, is used for described original image, carries out the image segmentation of M rows and N columns, obtains the image slice that M*N adjacent sides have L pixel overlapping, calculates the upper left corner of each image slice in the original image position coordinates;

[0088] A preprocessing module 903, configured to preprocess the M*N image slices;

[0089] The iterative segmentation module 904 is used to perform object recognition...

Embodiment 3

[0092] This embodiment also provides an adaptive level image segmentation recognition system, see Image 6 ,include:

[0093] One or more cameras 1 are used to record high-definition video of the scene; the camera uses a camera of at least 1080p to obtain high-definition scene images. The impact of 4k camera shooting is due to the high definition and the improvement of recognition results. The help will be greater, but the recognition speed will be slower than a 1080p camera.

[0094] One or more management servers 2 are used to execute the image acquisition step, the segmentation step and the iterative segmentation step, wherein,

[0095] The image acquisition step is to obtain the original image of a certain frame in the high-definition video; in this step, the camera communicates with the management server through a wired network or 5G wireless communication protocol, and transmits the real-time high-definition video to the management server. The image is sampled in the s...

technical field

[0001] The present invention relates to the technical field of high-definition image recognition, in particular to an adaptive hierarchical image segmentation recognition method, device and system. Background technique

[0002] Pedestrian recognition in densely populated places requires the use of high-definition cameras and object recognition technology, while traditional object recognition requires controlling image pixels in a relatively small size to improve recognition efficiency and adapt to the cache requirements of recognition devices. However, compressing images to a fixed size will result in the loss of a large number of small and medium object information, especially for densely populated areas such as train stations and squares. Using high-definition video acquisition equipment such as 4K and 8K resolution for monitoring, it is necessary to solve real-time and accurate object recognition for high-resolution images in video streams.

[0003] High-...

Examples

Embodiment 1

[0054] figure 1 It is a flow chart of an adaptive hierarchical image segmentation and recognition method provided by the embodiment of the present invention; an adaptive hierarchical image segmentation and recognition method provided by this embodiment will now be described in conjunction with an implementation example. The method includes the following steps:

[0055] Acquiring image step S101, obtaining the original image of a certain frame in the high-definition video;

[0056] Specifically, the video stream of a certain scene is captured by a high-definition camera (the resolution of the captured picture is at least 1920*1080), and the video stream data is transmitted to the memory of the management server through a wired network or a wireless network such as 5G. A frame of original image is sampled from the video stream at fixed intervals such as 0.1s. The size of the original image is 3840*2160. For specific images, see figure 2 .

[0057] Segmentation step S102, per...

Embodiment 2

[0085] Such as Figure 5 As shown, this embodiment provides an adaptive level image segmentation and recognition device, which is a virtual device corresponding to the adaptive level image segmentation and recognition method described in Embodiment 1, and the device is equipped to execute the The corresponding functional modules and beneficial effects of the method. The unit includes:

[0086] Obtaining an image module 901, configured to obtain an original image of a certain frame in the high-definition video;

[0087] Segmentation module 902, is used for described original image, carries out the image segmentation of M rows and N columns, obtains the image slice that M*N adjacent sides have L pixel overlapping, calculates the upper left corner of each image slice in the original image position coordinates;

[0088] A preprocessing module 903, configured to preprocess the M*N image slices;

[0089] The iterative segmentation module 904 is used to perform object recognition...

Embodiment 3

[0092] This embodiment also provides an adaptive level image segmentation recognition system, see Image 6 ,include:

[0093] One or more cameras 1 are used to record high-definition video of the scene; the camera uses a camera of at least 1080p to obtain high-definition scene images. The impact of 4k camera shooting is due to the high definition and the improvement of recognition results. The help will be greater, but the recognition speed will be slower than a 1080p camera.

[0094] One or more management servers 2 are used to execute the image acquisition step, the segmentation step and the iterative segmentation step, wherein,

[0095] The image acquisition step is to obtain the original image of a certain frame in the high-definition video; in this step, the camera communicates with the management server through a wired network or 5G wireless communication protocol, and transmits the real-time high-definition video to the management server. The image is sampled in the s...

Images