Agricultural greenhouse water and fertilizer management system and method based on video recognition

Abstract

The invention discloses an agricultural greenhouse water and fertilizer management system and method based on video recognition, and relates to the technical field of computer image recognition, and the system comprises a camera which is used for collecting crop images; the environment sensor is used for collecting water and fertilizer environment data of the agricultural greenhouse; the local host is provided with a crop identification model, and the crop identification model is used for identifying the real-time growth state of crops in the crop image; the cloud server is deployed with a growth decision model and a growth regulation model, the growth decision model formulates a growth control strategy according to the types and growth stages of crops in the agricultural greenhouse and the water and fertilizer environment data, the growth regulation model calculates an income value, and a regulation scheme for the growth control strategy is solved by taking the maximum income value as a target; and forming an adjusted growth control strategy. According to the method, the growth state of the crops is considered on the whole instead of achieving the optimal state of some crops, and the benefit of the agricultural greenhouse is improved.

Description

Technical Field

[0001] The present application relates to the field of computer image recognition technology, and in particular to an agricultural greenhouse water and fertilizer management system and method based on video recognition. Background Art

[0002] In modern agricultural production, greenhouses are a planting method that can achieve active control of the crop growth environment. Through automated lighting, temperature and humidity control and other means, it can create the most suitable environment for crop growth at any time, thereby greatly improving the quality and yield of agricultural products. Moreover, greenhouse production is almost unaffected by the season, so it has been widely used in agricultural production.

[0003] Agricultural greenhouses are typically operated over large areas, meaning multiple greenhouses are typically built within a single area. Therefore, a timely and efficient management system is required to manage multiple greenhouses simultaneously. Currently, greenhouse production management is based on the collection of various environmental data within the greenhouse, including images, light, temperature, humidity, and soil nutrients. Targeted environmental control is performed based on the crop's needs for light, temperature, humidity, and nutrients at different growth stages, in order to maintain optimal growth.

[0004] However, most current greenhouse environment management techniques are based on theoretical data determined through experience or experimentation, aiming to achieve theoretically optimal crop conditions by manipulating the environment. However, while these theoretical data have been extensively validated and proven to be feasible, greenhouses often grow multiple crops simultaneously. Even under optimal conditions, not all crops can achieve consistent, optimal conditions. This phenomenon is primarily due to two factors: First, the greenhouse environment is not uniform. Heat, moisture, air, and nutrients are unevenly distributed throughout the greenhouse, leading to varying degrees of variation in crop growth. Second, due to the influence of the crop's genetic makeup, even under optimal conditions, some genetically inferior crops may not achieve optimal conditions. Using existing management techniques that assume the same conditions for all locations and crops in a greenhouse will inevitably impact agricultural production returns. These environmental and genetic variations across locations exist objectively and cannot be eliminated, necessitating improvements to existing management techniques to optimize crop conditions and maximize returns. Summary of the Invention

[0005] The embodiments of the present application provide a video recognition-based agricultural greenhouse water and fertilizer management system and method to solve the problem in the prior art that the conditions of all crops in the greenhouse are the same, resulting in damage to agricultural production benefits.

[0006] On the one hand, the embodiment of the present application provides an agricultural greenhouse water and fertilizer management system based on video recognition, including: Cameras are installed at multiple locations in the agricultural greenhouse to collect crop images; Environmental sensors are installed in multiple locations in the agricultural greenhouse to collect water and fertilizer environmental data in the agricultural greenhouse; A local host is provided in the agricultural greenhouse and is in communication with the camera and the environmental sensor. The local host has a crop recognition model deployed therein. The crop recognition model is used to identify the real-time growth status of crops in the crop image. The real-time growth status includes the growth status level of the crops and the distribution area of the crops at each growth status level. A cloud server is set up in the cloud and communicates with the local host. A growth decision model and a growth regulation model are deployed in the cloud server. The growth decision model formulates a growth control strategy based on the type and growth stage of crops in the agricultural greenhouse, as well as water and fertilizer environment data. The growth regulation model calculates a benefit value according to different growth status levels and corresponding distribution areas. The benefit value represents the change in the growth status of all crops in the agricultural greenhouse relative to the real-time growth status after different adjustments are made to the growth control strategy. With the goal of maximizing the benefit value, an adjustment plan for the growth control strategy is solved to form an adjusted growth control strategy; An execution device executes the adjusted growth control strategy.

[0007] On the other hand, the embodiment of the present application also provides a method for managing water and fertilizer in an agricultural greenhouse based on video recognition, comprising: Collect crop images and water and fertilizer environment data in agricultural greenhouses; The crop recognition model is used to identify the real-time growth status of crops in the crop image. The real-time growth status includes the crop growth status level and the distribution area of crops at each growth status level. Develop growth control strategies based on crop types and growth stages as well as water and fertilizer environment data in agricultural greenhouses through growth decision models; The profit value is calculated according to different growth status levels and corresponding distribution areas. The profit value represents the change in the growth status of all crops in the agricultural greenhouse relative to the real-time growth status after making different adjustments to the growth control strategy; Taking the maximum benefit value as the goal, solve the adjustment plan of the growth control strategy and form the adjusted growth control strategy; Implement regulated growth control strategies.

[0008] The agricultural greenhouse water and fertilizer management system and method based on video recognition in this application has the following advantages: The growth status level of each crop is identified, and the growth status of each crop is predicted after adjusting the growth control strategy. The profit value is determined based on the change in growth status before and after the prediction. By solving the objective function of maximizing the profit value, a growth control strategy that maximizes the overall profit in the agricultural greenhouse can be obtained, rather than aiming to achieve the best state of certain crops. The growth status of crops is taken into consideration as a whole, which is conducive to improving the profit of the agricultural greenhouse. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0010] Figure 1 A schematic diagram of the composition of an agricultural greenhouse water and fertilizer management system based on video recognition provided in an embodiment of the present application.

[0011] Figure 2 A flowchart of a method for water and fertilizer management in an agricultural greenhouse based on video recognition is provided in an embodiment of the present application.

[0012] Description of the accompanying drawings: 100, camera; 200, environmental sensor; 300, local host; 400, cloud server; 500, execution device; 600, mobile terminal. DETAILED DESCRIPTION

[0013] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of the embodiments. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0014] Figure 1 This is a functional module diagram of an agricultural greenhouse water and fertilizer management system based on video recognition provided in an embodiment of the present application. This embodiment of the present application provides an agricultural greenhouse water and fertilizer management system based on video recognition, which includes: Cameras 100 are installed at multiple locations in the agricultural greenhouse and are used to collect crop images; Environmental sensors 200 are installed at multiple locations in the agricultural greenhouse. The environmental sensors 200 are used to collect water and fertilizer environmental data of the agricultural greenhouse; A local host 300 is provided in the agricultural greenhouse and is in communication with the camera 100 and the environmental sensor 200. The local host 300 has a crop recognition model deployed therein. The crop recognition model is used to identify the real-time growth status of crops in the crop image. The real-time growth status includes the crop growth status level and the distribution area of the crops at each growth status level. A cloud server 400 is provided in the cloud and is in communication with the local host 300. A growth decision model and a growth regulation model are deployed in the cloud server 400. The growth decision model formulates a growth control strategy based on the type and growth stage of crops in the agricultural greenhouse and water and fertilizer environment data. The growth regulation model calculates a benefit value according to different growth status levels and corresponding distribution areas. The benefit value represents the change in the growth status of all crops in the agricultural greenhouse relative to the real-time growth status after different adjustments are made to the growth control strategy. With the goal of maximizing the benefit value, an adjustment scheme for the growth control strategy is solved to form an adjusted growth control strategy; The execution device 500 executes the adjusted growth control strategy.

[0015] For example, multiple poles can be installed in the greenhouse, and the camera 100 can be placed at the top of each pole to facilitate the capture of crop images over the widest possible range. It should be understood that the camera 100 will capture real-time video from the agricultural greenhouse. After this video is sent to the local host 300, the local host 300 will capture frame images from the video at set time intervals as crop images.

[0016] In an embodiment of the present application, in order to cover the entire greenhouse, multiple cameras 100 need to be installed in the greenhouse, and these multiple cameras 100 are evenly distributed. For a single camera 100, although it can monitor a larger area, only a portion of the crop image it captures is clear. At a distance from the camera 100, the crop image often becomes blurred due to light, water mist, and other reasons. This portion of the image cannot be used in the crop recognition model. Therefore, after all cameras 100 are installed, the local host 300 can manually set the retention range of the crop image captured by each camera 100. Images outside the retention range will be cut off. After setting the retention range for all cameras 100, the combined retention ranges of all cameras 100 should just cover the entire greenhouse. The retention ranges of two adjacent cameras 100 should not overlap, but the boundaries of the retention ranges should also correspond to the same position in the greenhouse.

[0017] Environmental sensors 200 include a temperature and humidity sensor, a light sensor, and a water and fertilizer sensor. The temperature and humidity sensor further includes an air temperature and humidity sensor and a soil temperature and humidity sensor. The air temperature and humidity sensor is used to collect air temperature and humidity data at multiple locations in the greenhouse. The soil temperature and humidity sensor is inserted into multiple soil locations within the greenhouse to collect soil temperature and humidity data at multiple locations. The light sensor is used to collect light intensity at multiple locations within the greenhouse. The water and fertilizer sensor is inserted into multiple soil locations within the greenhouse to collect concentrations of specific substances, such as potassium, phosphorus, and nitrogen, at multiple locations. To facilitate deployment of environmental sensors 200, the air temperature and humidity sensor and the light sensor can be integrated together, while the soil temperature and humidity sensor and the water and fertilizer sensor can be integrated together.

[0018] The local host 300 can be set in a greenhouse, but in order to prevent the environment in the greenhouse from damaging the local host 300, an independent space can be built inside or outside the greenhouse to place the local host 300.

[0019] Furthermore, when identifying real-time growth status, the crop recognition model first separates each crop in the crop image and determines its location. It then identifies the shape and color of the leaves within the same crop and determines the crop's growth status based on these shapes and colors. Specifically, the purpose of separating each crop is to identify all leaves belonging to the same crop. Since leaves from the same crop tend to cluster together spatially and are connected to the same stem or branch, the location of the same crop can be identified through this spatial location and connection relationship.

[0020] After determining the leaves contained in each crop, the leaves can be further analyzed to obtain the morphology and color of the leaves. The morphology refers to the shape and flatness of the leaves, and the color is represented by RGB red, green and blue values.

[0021] After analyzing the shape and color of the leaves, the crop recognition model also obtains information about the type and growth stage of the crops in the current greenhouse. The type and growth stage are both obtained by the crop recognition model from the crop image. Based on this information, the standard leaf shape and color corresponding to the type and growth stage are obtained from the database. The identified leaf shape and color are compared with the standard leaf shape and color to obtain two differences. These differences are weighted and summed. The resulting value can represent the difference between the current leaf state and the standard state, so this value can be used as a basis for determining the growth state level. The growth state level in the embodiment of the present application is a set of several pre-set values. The larger the value, the higher the growth state level, and the smaller the value after the weighted sum of the corresponding two differences. Each growth state level value corresponds to a data range. When the value after the weighted sum of the two differences is within any data range, the leaf is in the growth state level corresponding to the data range. After determining the growth state level of all leaves in the same crop, the mode or median of all growth state levels is taken as the growth state level of the entire crop.

[0022] Furthermore, after determining the position of each crop, the crop recognition model aggregates the pixels belonging to the same crop in the crop image, with the number of pixels being the crop area. After determining the growth status level of each crop, the crop areas belonging to the same growth status level are summed to obtain the distribution areas of crops at multiple different growth status levels. Specifically, after separating each crop, all leaves of the same crop have been determined. Therefore, after determining the growth status level of the crop, the sum of the number of pixels of all leaves of the same crop in the crop image can be used as the crop area of the crop. After determining the growth status levels of all crops, the sum of the crop areas of all crops at the same growth status level is the distribution area of that growth status level.

[0023] It should be understood that although the number of pixels is affected by both the actual size and distance of the leaves, and cannot represent the actual leaf area, this situation exists for crops at all growth status levels. Therefore, using the number of pixels to represent the area is fair for all growth status levels.

[0024] Furthermore, the growth decision model obtains the corresponding control decision threshold according to the type and growth stage of crops in the agricultural greenhouse, and determines the adjustment parameters according to the difference between the control decision threshold and the water and fertilizer environment data. All the adjustment parameters constitute the growth control strategy.

[0025] Specifically, each crop has its own specific growth requirements at each growth stage. These growth requirements are reflected in the data as the requirements for light, air temperature and humidity, soil temperature and humidity, and water and fertilizer. In the laboratory, multiple experiments can be conducted on the same crop at each growth stage to determine the light, air temperature and humidity, soil temperature and humidity, and water and fertilizer data when the crop reaches the optimal growth state at each growth stage, and use these data as control decision thresholds. After the environmental sensor 200 collects the water and fertilizer environmental data at each moment, it calculates the difference between each water and fertilizer environmental data and the corresponding control decision threshold, and then generates adjustment parameters for controlling the operation of the execution device 500. When the execution device 500 operates according to the adjustment parameters, it adjusts the light, air temperature and humidity, soil temperature and humidity, and water and fertilizer in the greenhouse so that the environment inside the greenhouse can reach the control decision threshold, thereby providing good conditions for crop growth.

[0026] Furthermore, the growth regulation model first adjusts one or more regulation parameters in the growth control strategy, and each adjustment forms an adjustment plan. Then, based on the adjustment plan, the growth status of each crop in the future is predicted to obtain the predicted growth status. Then, the growth status level to which the predicted growth status belongs is determined, and the growth status level to which the real-time growth status of the same crop belongs is compared with the growth status level to which the predicted growth status belongs. The sub-benefit value of each crop is determined, and the sub-benefit values of all crops are summed up to obtain the benefit value.

[0027] Specifically, the growth control strategy formed above is only applicable when the environment at all locations in the greenhouse is consistent and the growth conditions of all crops are consistent. Obviously, this is unrealistic. There are natural differences in the environment and crops in the greenhouse. Such differences cannot be eliminated and can only be accepted. Based on such differences, the present application establishes an objective function for maximizing the benefit value. By solving the objective function, an adjustment scheme for adjusting the growth control strategy can be obtained. The adjustment scheme represents the adjustment size of each adjustment parameter in the growth control strategy. Therefore, after obtaining the adjustment scheme, the adjusted growth control strategy can be directly determined. However, there are many possibilities for adjusting the growth control strategy. There is only one growth control strategy that can meet the conditions for maximizing the benefit value. Therefore, a genetic algorithm can be used to solve the objective function and obtain the optimal adjustment scheme. In the solution process, the genetic algorithm can maximize the diversity of the adjustment scheme through selection, crossover and mutation, and then obtain the optimal adjustment scheme at the global level, avoiding falling into the local optimum.

[0028] After determining the adjusted growth control strategy based on the adjustment scheme, the growth status of the crop in the future can also be predicted. After the growth adjustment model forms a predicted growth state based on the adjusted growth control strategy, the growth state of the crop may get better or worse, which is reflected in the growth state level. The difference between the growth state level to which the predicted growth state belongs and the growth state level to which the real-time growth state belongs may be a positive number, a negative number, or of course zero. A positive number indicates that the predicted growth state is better than the current real-time growth state, a negative number indicates that the predicted growth state is worse than the current real-time growth state, and zero indicates that the predicted growth state is the same as the current real-time growth state. The purpose of adjusting the growth control strategy in this application is to make the predicted growth state better, or at least not worse than the current real-time growth state. Therefore, the difference between the growth state level to which the predicted growth state of each crop belongs and the growth state level to which the real-time growth state belongs is calculated to obtain a sub-benefit value. The sub-benefit values of all crops are weighted and summed to obtain the benefit value.

[0029] The weights used in the weighted summation of the above-mentioned sub-benefit values are determined according to the distribution area. Specifically, the distribution area represents the area of the crops at each growth state level, and the proportion of the distribution area at each growth state level in the distribution area of all crops in the greenhouse can be used as the weight for weighted summation. For a crop, no matter which growth state level its predicted growth state belongs to, the proportion of the distribution area of its growth state level corresponding to its real-time growth state in the distribution area of all crops in the greenhouse can be used as the weight of the sub-benefit value of the crop when participating in the weighted summation. If the distribution area of a certain growth state level accounts for a large proportion of the distribution area of all crops, it means that the growth state level occupies a dominant position in the greenhouse, so its influence on the benefit value is also greater. On the contrary, the growth state level that occupies a secondary position in the greenhouse has a smaller influence on the benefit value.

[0030] After each adjustment plan is obtained, the growth control strategy is adjusted according to the adjustment plan to obtain the corresponding intermediate strategy. The intermediate strategy is the adjusted growth control strategy. After applying the intermediate strategy, the growth adjustment model can predict the predicted growth status after a period of time in the future.

[0031] In a possible embodiment, the system of the present application also includes a mobile terminal 600, which is communicatively connected to the cloud server 400. The mobile terminal 600 is used to send control instructions to the cloud server 400, and the cloud server 400 forwards the control instructions to the local host 300, and finally the execution device 500 executes the control instructions.

[0032] Exemplarily, the mobile terminal 600 is an electronic device used by the manager of the greenhouse, which communicates with the cloud server 400 via a wireless network. The manager does not need to go to each greenhouse in person, but only needs to manually control the working status of each execution device 500 in the greenhouse at an appropriate time. Since each execution device 500 in this application preferentially works under the control of the local host 300, the manager can switch the working mode of the local host 300 on the mobile terminal 600. The working mode includes automatic mode and manual mode, wherein the automatic mode is a control program pre-deployed in the local host 300 that fully automatically controls the working status of each execution device 500. At this time, the local host 300 can forward all data generated during the control process to the mobile terminal 600 through the cloud server 400. If the manager believes that the control of the local host 300 is not appropriate, he can switch to manual mode and form corresponding control instructions by manually controlling each execution device 500.

[0033] Furthermore, after forming the adjusted growth control strategy, the cloud server 400 generates corresponding notification information and sends the notification information to the mobile terminal 600 .

[0034] The present application also provides a method for managing water and fertilizer in an agricultural greenhouse based on video recognition. Figure 2 As shown, the method includes the following steps: S200, collects crop images and water and fertilizer environment data in agricultural greenhouses; S210, identifying the real-time growth status of crops in the crop image using a crop recognition model, where the real-time growth status includes the crop growth status level and the distribution area of the crops at each growth status level; S220, formulating growth control strategies based on crop types and growth stages as well as water and fertilizer environment data in agricultural greenhouses through a growth decision model; S230, calculating a benefit value according to different growth status levels and corresponding distribution areas, where the benefit value represents a change in the growth status of all crops in the agricultural greenhouse relative to the real-time growth status after different adjustments are made to the growth control strategy; S240, with the goal of maximizing the benefit value, solving an adjustment scheme for the growth control strategy to form an adjusted growth control strategy; S250, executing the adjusted growth control strategy.

[0035] Although the preferred embodiments of the present application have been described, those skilled in the art may make additional changes and modifications to these embodiments once they have learned the basic creative concept. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present application.

[0036] Obviously, those skilled in the art may make various changes and modifications to this application without departing from the spirit and scope of this application. Thus, if these modifications and variations of this application fall within the scope of the claims of this application and their equivalents, this application is intended to include these modifications and variations.

Claims

1. An agricultural greenhouse water and fertilizer management system based on video recognition, characterized in that: include: Cameras (100) are arranged at multiple locations in the agricultural greenhouse, and the cameras (100) are used to collect crop images; Environmental sensors (200) are arranged at multiple locations in the agricultural greenhouse, and the environmental sensors (200) are used to collect water and fertilizer environmental data of the agricultural greenhouse; A local host (300) is arranged in an agricultural greenhouse, the local host (300) is in communication with the camera (100) and the environmental sensor (200), and a crop recognition model is deployed in the local host (300), the crop recognition model is used to recognize the real-time growth status of the crop in the crop image, the real-time growth status including the growth status level of the crop and the distribution area of the crop at each growth status level; A cloud server (400) is provided in the cloud, the cloud server (400) being in communication with the local host (300), wherein a growth decision model and a growth regulation model are deployed in the cloud server (400), wherein the growth decision model formulates a growth control strategy according to the type and growth stage of the crops in the agricultural greenhouse and the water and fertilizer environment data, and the growth regulation model calculates a benefit value according to different growth state levels and corresponding distribution areas, wherein the benefit value represents a change in the growth state of all crops in the agricultural greenhouse relative to the real-time growth state after different adjustments are made to the growth control strategy, and with the goal of maximizing the benefit value, an adjustment scheme for the growth control strategy is solved to form the adjusted growth control strategy; An execution device (500) executes the adjusted growth control strategy.

2. The agricultural greenhouse water and fertilizer management system based on video recognition according to claim 1 is characterized in that: The crop recognition model first separates each crop in the crop image and determines the position of each crop, then identifies the shape and color of the leaves of the same crop, and determines the growth status level of the crop based on the shape and color of the leaves.

3. The agricultural greenhouse water and fertilizer management system based on video recognition according to claim 2 is characterized in that: After determining the position of each crop, the crop recognition model aggregates the pixels belonging to the same crop in the crop image, and takes the number of pixels as the crop area. After determining the growth status level of each crop, the crop areas belonging to the same growth status level are summed to obtain the distribution areas of crops at multiple different growth status levels.

4. The agricultural greenhouse water and fertilizer management system based on video recognition according to claim 1 is characterized in that: The growth decision model obtains the corresponding control decision threshold according to the type and growth stage of the crops in the agricultural greenhouse, and determines the adjustment parameters according to the difference between the control decision threshold and the water and fertilizer environment data. All the adjustment parameters constitute the growth control strategy.

5. The agricultural greenhouse water and fertilizer management system based on video recognition according to claim 4 is characterized in that: The type and the growth stage are both obtained by identifying the crop image through the crop recognition model.

6. The agricultural greenhouse water and fertilizer management system based on video recognition according to claim 4 is characterized in that: The growth regulation model first adjusts one or more of the regulation parameters in the growth control strategy, and each adjustment forms an regulation scheme. Then, based on the regulation scheme, the growth status of each crop in the future is predicted to obtain a predicted growth status. Then, the growth status level to which the predicted growth status belongs is determined, and the growth status level to which the real-time growth status of the same crop belongs is compared with the growth status level to which the predicted growth status belongs to determine the sub-benefit value of each crop. The sub-benefit values of all crops are summed to obtain the benefit value.

7. The agricultural greenhouse water and fertilizer management system based on video recognition according to claim 6 is characterized in that: After each adjustment scheme is obtained, the growth control strategy is adjusted according to the adjustment scheme to obtain a corresponding intermediate strategy. After applying the intermediate strategy, the predicted growth state after a period of time in the future is predicted.

8. The agricultural greenhouse water and fertilizer management system based on video recognition according to claim 1 is characterized in that: The system further includes a mobile terminal (600), wherein the mobile terminal (600) is in communication with the cloud server (400), and the mobile terminal (600) is used to send a control instruction to the cloud server (400), and the cloud server (400) forwards the control instruction to the local host (300), and finally the execution device (500) executes the control instruction.

9. The agricultural greenhouse water and fertilizer management system based on video recognition according to claim 8, characterized in that: After forming the adjusted growth control strategy, the cloud server (400) generates corresponding notification information and sends the notification information to the mobile terminal (600).

10. A method for an agricultural greenhouse water and fertilizer management system based on video recognition as described in any one of claims 1 to 9, characterized in that: include: Collect crop images and water and fertilizer environment data in agricultural greenhouses; identifying a real-time growth status of crops in the crop image using a crop recognition model, wherein the real-time growth status includes a growth status level of the crops and a distribution area of the crops at each growth status level; formulating a growth control strategy according to the type and growth stage of the crops in the agricultural greenhouse and the water and fertilizer environment data through a growth decision model; Calculating a benefit value according to different growth status levels and corresponding distribution areas, the benefit value representing a change in the growth status of all crops in the agricultural greenhouse relative to the real-time growth status after making different adjustments to the growth control strategy; Taking the maximum benefit value as the goal, solving an adjustment scheme for the growth control strategy to form the adjusted growth control strategy; The adjusted growth control strategy is implemented.

Images