feat: 🚧 Continue technical part redaction

This commit is contained in:
2024-10-31 16:06:27 +01:00
parent d515d62c1e
commit aa45331010
13 changed files with 358 additions and 6 deletions

View File

View File

@ -0,0 +1 @@
Currently, we have not yet started implementing the OCR. For the first project review, we initially aimed to deliver a functional OCR for simple images. However, we realized that this part of the project is very challenging. We have therefore decided to delay its start, focusing first on completing the image processing functions and the initial AI algorithms. This will allow us to fully concentrate on the OCR afterward.

View File

@ -1,2 +1,11 @@
This section aims to explain each part of the project from a technical perspective. It includes the libraries used, the various applications related to the project, how the project is built, and more.
\subsection{Traitement de l'image}
\input{sections/partie-technique/traitement-image/traitement-image}
\subsection{OCR}
\input{sections/partie-technique/OCR/ocr.tex}
\subsection{Artificial Intelligence}
\input{sections/partie-technique/IA/IA.tex}

View File

@ -0,0 +1,5 @@
The second step of the process involves converting each pixel from grayscale to black and white. This further simplifies the image by retaining only two colors: black and white. As a result, the OCR can more effectively recognize different characters and process the information more quickly, since it only has to differentiate between two color values: black and white.
To complete this process, the functions \verb'void pixel_to_black_white' and \verb'void pixel_to_grayscale' are executed sequentially. This results in an output image in grayscale.
TODO: Add some images to illustrate the process.

View File

@ -0,0 +1,3 @@
La deuxième partie du processus consiste à transformer chaque pixel en nuance de gris en pixel black and white. Cela permet de simplifier encore plus l'image en ne gardant que deux couleurs, le blanc et le noir. Cela permet ensuite à l'OCR de mieux reconnaître les différentes lettres et de traiter les informations plus rapidement car il n'a plus que deux informations de couleurs, soit du noir, soit du blanc.
Cette conversion est assuré par la fonction \verb'void pixel_to_black_white'. Elle prends en paramètre un \verb|void* pixel|, \verb|SDL_PixelFormat* format| et \verb|Uint8 threshold|. La fonction applique le traitement directement sur l'image.

View File

@ -0,0 +1,3 @@
The first step of the process involves converting each pixel to grayscale. This simplification removes most color information from the image, enabling the OCR to better recognize different characters and process the information more quickly since it no longer has to deal with color data.
This conversion is handled by the function \verb'void pixel_to_grayscale'. It takes as parameters a \verb|void* pixel| and an \verb|SDL_PixelFormat*format|, applying the necessary processing to achieve the grayscale effect.

View File

@ -0,0 +1,10 @@
Cette sous section détaille le processus de traitement d'image utilisé dans notre logiciel, afin de la préparer pour qu'elle soit analyser de la meilleure des façons par l'OCR\footnote{cf. section X.X sur l'OCR}. Pour ce faire, nous faisons passer l'image par plusieurs étapes avant de l'envoyer à l'OCR.
\subsubsection{Transformer les pixels en nuances de gris}
\input{sections/partie-technique/traitement-image/pixel-grayscale/pixel-grayscale}
\subsubsection{Transformer les pixels en blank and white}
\input{sections/partie-technique/traitement-image/pixel-baw/pixel-baw}
\subsubsection{Transformer l'image en nuance de gris}
\input{sections/partie-technique/traitement-image/image-baw/image-baw}