The software based on internet environment has three basic features:

• The database of raw images of BSL to be downloaded and uploaded;
• The interactive environment to analyse the BSL data by the user online;
• The Tutorial area for the users to obtain more information about the BSL.

With these features, the users can develop their research by means of raw data from research already published, thus recognised by third parties. The exchange of information by the data-packs also offers the possibility of new joint works and the enhancement of the technique around the world.

Using the interactive function to address on-line the graphical and the numerical information from their own data, the users feel safe using the technique, since some users do not have sufficient knowledge regarding image mathematical computer processing. Finally, the test of data from others, obtained in the portal, will help the users to feel confident when testing their own data.

The designed software can be summarised in the following Metadata and can be divided in three main modules: Accessing and building a database and tutorials; Administrative features; and Data analysis.

Metadata

Current code version: V.1.0.1 Code archived DOI: 10.5281/zenodo.10038616 (Godinho 2023)

Link to code/repository used for this code version: https://github.com/robsonpg/Bslonclouds

Link to reproducible capsule: https://bslonclouds.com/

Legal code licence: GNU General Public Licence (GPL) v.3

Code versioning system used: git

Software code languages, tools and services used: PHP, Javascript

Support email for questions: contact@bslonclouds.com

Accessing and building a database and tutorials

The two first features were obtained using the PHP language, since it is a language that is accepted in the majority of internet service hosts, including the low-cost servers. The backend build was programmed to receive the upload of the images, submitted to tests of integrity and receiving a moderation control before release to public access.

The managing of the images was provided by Javascript that runs in the user’s machine. The distributed processing is the key-factor of the efficiency of the procedures (Fig. 1). However, the seeking of efficiency of image management is done in other procedures, such as in the image upload. For instance, the upload of the images occurs with them placed in a line to be sent to the backend. The line is done to avoid the overload of the web connection and to avoid time-consuming in the backend.

The images are stored in a MySQL database and are linked to the additional information related to the image generation (metadata), such as information about the laser, the camera, the time-rate of image assembling and other relevant data to allow their use by third parties. The owner of the images can also include the DOI or the URL of where they were published originally, paper or report.

When the images are ready to start uploading, a report with all the information (metadata) is created and approved by the owner. The images are sent one-by-one to avoid overloading the backend. Thus, when the user enables the uploading, the metadata and the sequence of images are received in the backend to be released by the moderator, who checks the pertinence of the data.

Once approved to publish by the website administrator, the data are placed in an environment named Catalog, where the users can find all the collection of data to proceed with the download.

The gateway was built upon a framework with protection during each step of the use. This protocol aimed to achieve a fast process during the reception of the images comprising the analysis of the files and their assembling.

Administrative features of the solution

In the administrative feature of the solution, the security of the data and the control of access were implemented, with black and white IP list checking. The pages are individually treated receiving permissions of access to avoid the irregular use of the system’s rules by common users. In addition, regarding the security, the portal uses the protocol HTTPS.

The management of users is already provided for the administrators of the portal, with options of use permissions for some functionalities or even exclusion. The system also provided a link of contact by means of an email.

A backup function of the data in the portal was built and the operation of the system is registered and stored.

Data analysis

The online analysis of data from users is provided by JavaScript running graphical and numerical functions as presented in Equation 1 and Equation 2, respectively.

\(GAVD = E[|I_k - I_{k+1}|]\) (1)

where GAVD is an outcome matrix representing the heat map in grey or pseudocolours, relative to the map of activity and the I is the image in time k, from 1 to N (number of images acquired in a particular time rate in seconds); E[.] is the expectation operator [14]. The expectation is obtained from the occurrence matrix [14], that is a secondary matrix, linked to the time history of speckle patterns (THSP) of each pixel of the images in time. Thus, the THSP is a matrix with the values of the selected pixels in time.

The GAVD is, therefore, the graphical outcome of the Absolute Value of the Differences (AVD) function, that is used in the numerical analysis of the BSL. In the online tool, we used the AVD#1 represented by the Equation 2

\(AVD = E [| i - j |]\) (2)

where i and j are the grey values in the occurrence matrix [14]. In the numerical approach, the THSP is used to construct the Co-Occurrence Matrix, and the THSP is formed by random points, in a Gaussian distribution, collected from an area selected by the user interactively.

In our interactive tool, we used the graphical and numerical outcomes to help the user to test the technique and to have easy contact with it, which allows the learning of the graphical and numerical features and their meaning. The human-machine interface can be summarised by some steps.

In Fig. 2, we can observe the first step linked for the preparation of the dataset, here named as datapack. Thus, a collection of RGB images in 2D matrices is transformed in grey-scale, when it is the case and, thus, a sequence of images is created forming a 3D matrix.

The second step is the generation of the map of activity from the sequence of speckle pattern images. Using Javascript’s features, we dealt with the images, using them as an up-dated buffer for the mathematical operations. In addition, to reduce the time-consuming and increase the security of the operation, the developed programme was run in the user’s local machine, thus avoiding the traffic of data. In Fig. 3, we can see the process of image checks, size and number of images, followed by the image processing using the interactive AVD function.

In the numerical process of the AVD, the operation of matrices was replaced by images manipulation, using the best of Javascript. The images (matrices with grey-scale values) were used to aid the user to define the desired area for the numerical analysis (with a graphical outcome) and as the datapack of the numerical analysis.

The data used in the numerical analysis is constructed from a collection of random points weighted in a Gaussian distribution. The interactive operation allows the tailoring of the Gaussian function (the number points and the standard deviation) by the user, as well as the selection of the spot area where the numerical analysis will be conducted. A graphical outcome of AVD is plotted to aid the user to move the selected points using a computer mouse over the graphical map of activity (graphical AVD outcome).

Therefore, the microprocessor responsible for the operations, such as the rendering, of the images is the graphic processor, which, in most hardware, is represented by a dedicated graphics card. While the memories of the mother-boards commonly use the DDR4 (~3,200 MHz), the dedicated graphic cards can use the DDR6 (12,500 MHz).

However, when the user’s computer does not have a dedicated graphic card, the system reserves an area of the RAM memory for image renderisation. In this case, the running of the online analysis will take some seconds more, but not compromising the online sensation.

The distributed operation associated with the image processing in Javascript offers an experience of real-time analysis, when the graphical outcome can be used interactively by the user to obtain the numerical value of AVD within the random-gaussian points positioned by the user him/herself. Therefore, the numerical values can be accessed in real-time and the user can evaluate in the whole image the biological activity measured by the AVD. This procedure can be run as many times as the users wish over the graphical outcome as a guide.

An option of off-line analysis is offered to those needing further and detailed analysis.

The functionality of dealing with raw data can be seen in Fig. 5, where a crop of the download screen presents the view of the data with its thumbnail and all descriptions of the data, such as the sample illuminated, the type of laser, time rate of image assembling etc. The data are downloaded in “zip” format, that includes the sequence of images and the metadata.

In Fig. 6, we can observe the outcome of the Online Analysis function with the result of a maize seed analysis. The outcome presents the graphical and the numerical results, where the graphical is a low resolution preview of the map of activity. The preview image is the reference to the user selecting with the mouse the area where she/he wants to evaluate the numerical outcome, the activity presented by the AVD.

To guide the user, a histogram of the points is presented, helping the selection of an area without saturation or under-exposure of light. The area and the number of points in Gaussian distribution can be tailored by the user in order to sharply define the observed area, i.e. avoiding the mixing of an area with high activity (seed) with another with low activity.