A high-definition adaptive video surveillance system

Abstract

The present application discloses a high-definition adaptive video surveillance system, which can automatically adjust the parameters of video surveillance according to changes in the monitoring environment to ensure that high-definition, high-quality monitoring images are always obtained. By adopting a unique adaptive algorithm, key indicators such as image brightness, contrast, and color are analyzed and adjusted in real time, and combined with scene recognition technology, different monitoring modes are automatically switched, which greatly improves the adaptability and reliability of the monitoring system, and can be widely used in security monitoring, traffic monitoring, industrial monitoring and other fields.

Description

Technical Field

[0001] The present application relates to the field of video surveillance technology, and in particular to a high-definition adaptive video surveillance system. Background Art

[0002] With the development of society and the advancement of science and technology, video surveillance systems have been widely used in various fields. However, traditional video surveillance systems have some limitations. For example, it is often difficult to obtain clear and stable surveillance images under different lighting conditions, weather conditions and complex scenes. In order to overcome these problems, a high-definition video surveillance system that can automatically adapt to environmental changes is needed to improve the effect and quality of surveillance. Summary of the invention

[0003] The main purpose of the present invention is to provide a high-definition video monitoring system that can automatically adapt to environmental changes to improve the monitoring effect and quality.

[0004] To achieve the above purpose, this high-definition adaptive video surveillance system is mainly composed of an image acquisition module, an image processing module, a control module, a storage module and a display module. Among them, the image acquisition module is responsible for acquiring video images of the monitoring site, including a camera and its related optical components and sensors; the image processing module is used to analyze and process the acquired images, and use an adaptive algorithm to optimize the image quality; the control module automatically adjusts the parameters of the image acquisition module according to the analysis results of the image processing module, such as exposure time, aperture size, gain, etc.; the storage module is used to save the monitoring video and related data information; the display module displays the processed high-definition images to the user.

[0005] Preferably, the adaptive algorithm includes a brightness adaptive algorithm, a contrast adaptive algorithm, a color adaptive algorithm and a scene recognition and mode switching algorithm.

[0006] Preferably, the brightness adaptive algorithm includes the following steps: grayscale the captured image, calculate the average brightness value of the grayscale image, determine whether the brightness of the current image is within an appropriate range based on a preset brightness threshold, and if not, increase or decrease the exposure time or gain accordingly, and set the adjustment step to avoid over-adjustment.

[0007] Preferably, the contrast adaptive algorithm includes calculating the contrast of the image, judging whether the contrast meets the requirement according to a preset contrast threshold, and if not, enhancing the contrast of the image by methods such as histogram equalization.

[0008] Preferably, the color adaptation algorithm includes converting the image from the RGB color space to the HSV color space, calculating the average values ​​of the hue, saturation and brightness of the image, and adjusting the color of the image by adjusting the color correction matrix according to a preset target value.

[0009] Preferably, the scene recognition and mode switching algorithm includes using a deep learning algorithm to perform scene recognition on the collected images, automatically switching to a corresponding monitoring mode according to the recognition results, and adjusting the parameters of the image acquisition module.

[0010] The high-definition adaptive video monitoring system provided by the present invention:

[0011] 1. Improve image quality: Through the adaptive algorithm, the brightness, contrast, color, etc. of the image can be adjusted in real time, so that clear and high-quality monitoring images can be obtained under different environmental conditions, which greatly improves the monitoring effect and reliability.

[0012] 2. Enhance system adaptability: The system can automatically identify different monitoring scenarios and switch to the corresponding monitoring mode without manual intervention, adapting to complex and changing monitoring environments and reducing maintenance costs.

[0013] 3. Broad application prospects: The present invention can be widely used in security monitoring, traffic monitoring, industrial monitoring and other fields, and is of great significance for ensuring social security and improving production efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the structures shown in these drawings without paying creative work.

[0015] Figure 1 This is a principle block diagram of a high-definition adaptive video surveillance system in an embodiment of the present application;

[0016] Figure 2 is a flow chart of a brightness adaptive algorithm in an embodiment of the present application;

[0017] Figure 3 is a flow chart of a contrast adaptive algorithm in an embodiment of the present application;

[0018] Figure 4 is a flow chart of a color adaptive algorithm in an embodiment of the present application;

[0019] Figure 5 is a flow chart of a scene recognition and mode switching algorithm in an embodiment of the present application;

[0020] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0021] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0022] It should be noted that all directional indications in the embodiments of the present invention (such as up, down, left, right, front, back, etc.) are only used to explain the relative position relationship, movement status, etc. between the components under a certain specific posture (as shown in the accompanying drawings). If the specific posture changes, the directional indication will also change accordingly.

[0023] In addition, the descriptions of "first", "second", etc. in the present invention are only used for descriptive purposes and cannot be understood as indicating or implying their relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include at least one of the features. In addition, the technical solutions between the various embodiments can be combined with each other, but they must be based on the ability of ordinary technicians in the field to implement them. When the combination of technical solutions is contradictory or cannot be implemented, it should be deemed that such a combination of technical solutions does not exist and is not within the scope of protection required by the present invention.

[0024] In one embodiment, if Figure 1 As shown, the present application discloses a high-definition adaptive video surveillance system, which is mainly composed of an image acquisition module, an image processing module, a control module, a storage module and a display module. Among them, the image acquisition module is responsible for acquiring video images of the monitoring scene, including a camera and its related optical components and sensors; the image processing module is used to analyze and process the acquired images, and use an adaptive algorithm to optimize the image quality; the control module automatically adjusts the parameters of the image acquisition module according to the analysis results of the image processing module, such as exposure time, aperture size, gain, etc.; the storage module is used to save monitoring videos and related data information; and the display module displays the processed high-definition images to the user. Its framework is shown below:

[0025] A[image acquisition module]-->B[image processing module];

[0026] B-->C[control module];

[0027] C-->A;

[0028] B-->D[storage module];

[0029] B-->E[display module];

[0030] Image preprocessing: Get the original image of the monitoring scene (which can be a color image) from the image acquisition module, and scale it to the input size required by the target detection network (for example, scale images of different resolutions to a size of 608 pixels in width and 608 pixels in height, depending on the selected network structure).

[0031] Normalize the scaled image by subtracting the mean from the RGB value of each pixel and dividing it by the standard deviation (the mean and standard deviation can be determined based on the dataset used for network pre-training. For example, the mean and standard deviation of the commonly used ImageNet dataset are [0.485, 0.456,

[0032] 0.406] and [0.229, 0.224, 0.225]), so that the image data conforms to the input data distribution during network training, which facilitates the network to perform accurate feature extraction and reasoning.

[0033] Feature extraction and target prediction: The preprocessed image is input into a pre-trained deep convolutional neural network (such as the improved version of the YOLOv5 network). The network gradually performs deep feature extraction on the image through multiple convolutional layers, pooling layers, residual structures and other components to obtain feature maps of different scales (for example, in YOLOv5, multiple feature maps of different resolutions are generated to detect targets of different sizes).

[0034] Based on the extracted feature map, the detection head in the network uses convolution operations to predict the category probability of the target (such as the confidence scores of various categories such as people, cars, objects, etc.), the location coordinates of the target (expressed in the form of a bounding box, usually the coordinates of the upper left and lower right corners of the bounding box or the coordinates of the center point plus the width and height information), and the approximate size information of the target (which can be inferred from the size of the bounding box), and outputs the corresponding prediction result tensor.

[0035] Non-maximum suppression (NMS) processing: Since the target detection network may generate multiple overlapping predicted bounding boxes for the same target, non-maximum suppression processing is required to screen out the optimal target bounding box.

[0036] Traverse the predicted bounding boxes under each category, sort them according to the category confidence score, start with the bounding box with the highest score, and calculate its intersection over union (IoU, that is, the ratio of the overlapping area of ​​two bounding boxes to the union area) with other bounding boxes of the same category.

[0037] If the IoU of a bounding box of the same category as the current highest-score bounding box is greater than a preset IoU threshold (e.g., IoU_threshold = 0.5), then this bounding box is considered redundant and is deleted. Only the optimal bounding box corresponding to each target and its category, position, and size information are retained to obtain the final accurate object detection result, which is passed to the subsequent scale adaptive adjustment strategy module.

[0038] Calculation of target distance and proportion of the screen occupied: According to the target position coordinate information output by the object detection network and in combination with the distance sensor data in the environmental perception module, calculate the relative distance distance between the target object and the camera (for example, by combining the principle of triangulation ranging with the approximate distance range of the target feedback by the distance sensor to further accurately calculate the actual distance from the target to the camera. The specific ranging algorithm depends on the sensor type and system design).

[0039] At the same time, calculate the area ratio area_ratio of the target object in the screen by calculating the width and height of the target bounding box and the width and height of the entire screen. The formula can be expressed as: area_ratio = (box_width * box_height) / (image_width * image_height) (where box_width and box_height are the width and height of the target bounding box, and image_width and image_height are the width and height of the monitoring screen image).

[0040] Parameter adjustment decision: If the target is relatively close to the camera (e.g., distance < close_threshold, where close_threshold is a preset close-distance threshold that can be determined through experiments) and the proportion of the target in the screen is too large (e.g., area_ratio > large_ratio_threshold, where large_ratio_threshold is also a threshold set according to actual requirements), then it is determined that the focal length needs to be reduced and the resolution allocation needs to be decreased. The specific operation is to send an instruction to the image acquisition module through the parameter control module to adjust the focal length and resolution parameters of the camera according to the set adjustment step (e.g., the focal length is reduced by a certain angle value each time, and the resolution is reduced to the next appropriate level, etc.) to reduce the data volume and improve the system processing efficiency while ensuring that the key details of the target can be clearly seen.

[0041] Conversely, when the target distance is far (e.g., distance > far_threshold) and the proportion in the picture is small (e.g., area_ratio < small_ratio_threshold), an operation to increase the focal length and improve the resolution of the corresponding area is triggered. Moreover, a super-resolution reconstruction algorithm is called (for example, the image of the target area to be enhanced is cropped out and input into a super-resolution algorithm based on a generative adversarial network (GAN). The generator generates a high-resolution image, and the discriminator judges the difference from the real high-resolution image for adversarial training to continuously optimize the generation effect) to perform detail enhancement processing on the distant target area, so that even small distant targets can clearly show their key features.

[0042] When the target distance and the proportion in the picture are in an intermediate state, the current camera parameters are maintained unchanged, and the change of the target is continuously monitored for timely adjustment, so as to realize the adaptive adjustment of parameters according to the actual scale of the target and ensure the high-definition monitoring effect of the target in the whole scene.

[0043] Furthermore, the adaptive algorithm includes a brightness adaptive algorithm, a contrast adaptive algorithm, a color adaptive algorithm, and a scene recognition and mode switching algorithm.

[0044] Among them,

[0045] Referring to Figure 2 , the brightness adaptive algorithm includes the following steps: grayscale the collected image, calculate the average brightness value of the grayscale image, judge whether the brightness of the current image is within a suitable range according to a preset brightness threshold. If it is not within the range, increase or decrease the exposure time or gain accordingly, and set an adjustment step size to avoid over-adjustment. Its process is as follows:

[0046] A [Collect image] --> B [Grayscale processing];

[0047] B --> C [Calculate average brightness value];

[0048] C --> D {Is brightness lower than threshold?};

[0049] D --> |Yes| E [Increase exposure time / increase gain];

[0050] D --> |No| F {Is brightness higher than threshold?};

[0051] F --> |Yes| G [Decrease exposure time / decrease gain];

[0052] F --> |No| H [Output image];

[0053] E --> H;

[0054] G -> H;

[0055] First, the collected image is grayed out to convert the color image into a grayscale image.

[0056] In order to analyze the brightness information more accurately.

[0057] Calculate the average brightness value L of the grayscale image aug .

[0058] According to the preset brightness threshold T low and T high The brightness of the current image is determined to determine whether the brightness of the current image is within an appropriate range.

[0059] If L aug <T low , increase the exposure time or increase the gain to increase the brightness of the image; conversely, if L aug >T high , reduce the exposure time or lower the gain.

[0060] In order to avoid image quality degradation caused by excessive adjustment, the adjustment step S is set so that the adjustment amplitude does not exceed S each time.

[0061] Further, refer to Figure 3 The contrast adaptive algorithm includes calculating the contrast of the image, judging whether the contrast meets the requirement according to a preset contrast threshold, and if not, enhancing the contrast of the image by methods such as histogram equalization.

[0062] The process is as follows:

[0063] A[Acquired image]-->B[Grayscale processing];

[0064] B-->C[calculate the average brightness value];

[0065] C-->D{Is the brightness lower than the threshold?};

[0066] D-->|Yes|E[Increase exposure time / increase gain];

[0067] D-->|No|F{Is the brightness higher than the threshold};

[0068] F-->|Yes|G[reduce exposure time / reduce gain];

[0069] F-->|No|H[output image];

[0070] E-->H;

[0071] G -> H;

[0072] Calculate the contrast C of the image. The contrast calculation formula is: Where Lmax and Lmin are the maximum and minimum luminance values in the image, respectively.

[0073] Judge whether the contrast of the current image meets the requirements according to a preset contrast threshold.

[0074] If C < Tc, enhance the contrast of the image by methods such as histogram equalization to make the details of the image more clearly visible.

[0075] Furthermore, referring to Figure 4 , the color adaptive algorithm includes converting the image from the RGB color space to the HSV color space, calculating the average values of the hue, saturation, and lightness of the image, and adjusting the color of the image by adjusting the color correction matrix according to preset target values.

[0076] Convert the image from the RGB color space to the HSV color space to facilitate the processing of color information more conveniently.

[0077] Calculate the average values of the hue H, saturation S, and lightness V of the image respectively.

[0078] Set the target values Ht, St, and Vt of the hue, saturation, and lightness according to different monitoring scenarios and user requirements.

[0079] Adjust the color of the image to be close to the target value by adjusting the color correction matrix to make the color of the image more real and natural.

[0080] Furthermore, referring to Figure 5 , the scene recognition and mode switching algorithm includes using a deep learning algorithm to perform scene recognition on the collected image, automatically switching to the corresponding monitoring mode according to the recognition result, and adjusting the parameters of the image acquisition module. The process is as follows:

[0081] A [Collect image] --> B [Scene recognition];

[0082] B --> C {Scene type judgment};

[0083] C --> |Daytime| D [Daytime mode];

[0084] C --> |Night| E [Night mode];

[0085] C --> |Strong light| F [Strong light mode];

[0086] C --> |Weak light| G [Weak light mode];

[0087] D --> H [Adjust parameters and monitor];

[0088] E --> I [Turn on infrared fill light, adjust parameters and monitor];

[0089] F-->J[reduce exposure, increase contrast and monitor];

[0090] G-->K[adjust parameters and monitor];

[0091] Use deep learning algorithms to perform scene recognition on the collected images and identify the current monitoring scene as daytime, nighttime, indoor, outdoor, strong light, weak light, etc. According to the results of scene recognition, the system automatically switches to the corresponding monitoring mode. For example, in night mode, the system automatically turns on the infrared fill light function and adjusts the image parameters to adapt to low light conditions; in strong light mode, it reduces the exposure and enhances the contrast to prevent overexposure of the image.

[0092] In actual application, when the monitoring system is started, the image acquisition module starts to collect video images of the monitoring site in real time and transmits the image data to the image processing module. The image processing module first analyzes the brightness, contrast and color of the image, calculates the corresponding adjustment parameters using the above-mentioned adaptive algorithm, and transmits the parameters to the control module. According to the received parameters, the control module automatically adjusts the exposure time, aperture size, gain and other parameters of the image acquisition module to optimize the image quality. At the same time, the scene recognition and mode switching algorithm will continuously perform scene recognition on the collected images, automatically switch the monitoring mode according to different scenes, and ensure that the system can obtain the best monitoring effect in various environments. The storage module will regularly save the monitoring video and related data information for subsequent query and analysis. The display module displays the processed high-definition monitoring images in real time for users to view.

[0093] The above description is only a preferred embodiment of the present invention, and does not limit the patent scope of the present invention. All equivalent structural changes made by using the contents of the present invention specification and drawings under the inventive concept of the present invention, or directly / indirectly applied in other related technical fields are included in the patent protection scope of the present invention.

Claims

1. A high-definition adaptive video surveillance system, characterized in that: It includes an image acquisition module, an image processing module, a control module, a storage module and a display module. The image acquisition module acquires video images of the monitoring site and transmits them to the image processing module. The image processing module uses an adaptive algorithm to analyze and process the image and transmits the processing result to the control module. The control module automatically adjusts the parameters of the image acquisition module according to the processing result. The storage module is used to save the monitoring video and related data information. The display module is used to display the processed high-definition image.

2. A high-definition adaptive video surveillance system according to claim 1, characterized in that: The adaptive algorithms include a brightness adaptive algorithm, a contrast adaptive algorithm, a color adaptive algorithm, and a scene recognition and mode switching algorithm.

3. A high-definition adaptive video surveillance system according to claim 2, characterized in that: The brightness adaptive algorithm includes the following steps: graying the captured image, calculating the average brightness value of the grayscale image, judging whether the brightness of the current image is within an appropriate range according to a preset brightness threshold, and if not within the range, increasing or decreasing the exposure time or gain accordingly, and setting the adjustment step to avoid over-adjustment.

4. A high-definition adaptive video surveillance system according to claim 2, characterized in that: The contrast adaptive algorithm includes calculating the contrast of the image, judging whether the contrast meets the requirement according to a preset contrast threshold, and if not, enhancing the contrast of the image by methods such as histogram equalization.

5. The high-definition adaptive video surveillance system according to claim 2, characterized in that: The color adaptation algorithm includes converting the image from the RGB color space to the HSV color space, calculating the average values ​​of the hue, saturation and brightness of the image, and adjusting the color of the image by adjusting the color correction matrix according to a preset target value.

6. A high-definition adaptive video surveillance system according to claim 2, characterized in that: The scene recognition and mode switching algorithm includes using a deep learning algorithm to perform scene recognition on the collected images, automatically switching to the corresponding monitoring mode according to the recognition results, and adjusting the parameters of the image acquisition module.

Images