In this step we have to detect all the plates in the current camera frame. To do this task, we divide it in two main steps: segmentation and segment classification. The feature step is not explained because we use the image patch as a vector feature.
In the first step (segmentation), we apply different filters, morphological operations, contour algorithms, and validations to retrieve those parts of the image that could have a plate.
In the second step (classification), we apply a Support Vector Machine (SVM) classifier to each image patch—our feature. Before creating our main application we train with two different classes—plate and non-plate. We work with parallel frontal-view color images that are 800 pixels wide and taken 2–4 meters from a car. These requirements are important to ensure correct segmentations. We can perform detection if we create a multi-scale image algorithm.
In the next image we have shown all the processes involved in plate detection: