FIELD: image data processing.
SUBSTANCE: invention relates to document recognition systems. The system and method for recognition of documents in an image include creating a superpixel segmentation map of the initial image using a neural network; generating a superpixel binary mask by linking each superpixel from the segmentation map with a class from a preset class set; identifying one or more connected components in the superpixel binary mask; identifying the corresponding minimum circumscribing polygon for each connected component of the binary mask; creating one or more lines dividing the image based on the minimal circumscribing polygons; and determining the boundaries of one or more objects of interest based on at least a subset of the lines dividing the image.
EFFECT: expanded range of means for the purpose of document recognition.
20 cl, 7 dwg
| Title | Year | Author | Number |
|---|---|---|---|
| IMAGE PROCESSING SYSTEM AND METHOD USING DEEP NEURAL NETWORKS | 2018 |
|
RU2743931C1 |
| TRAINING GAN (GENERATIVE ADVERSARIAL NETWORKS) TO CREATE PIXEL-BY-PIXEL ANNOTATION | 2019 |
|
RU2735148C1 |
| DETECTING AND IDENTIFYING OBJECTS ON IMAGES | 2020 |
|
RU2726185C1 |
| SOFTWARE AND HARDWARE COMPLEX DESIGNED FOR PROCESSING AEROSPACE IMAGE OF TERRAIN FOR PURPOSE OF DETECTION, LOCALIZATION AND CLASSIFICATION BY TYPE OF AVIATION AND LAND EQUIPMENT | 2021 |
|
RU2811357C2 |
| METHOD FOR RECOGNITION OF TEXT IN IMAGES OF DOCUMENTS | 2021 |
|
RU2768544C1 |
| METHOD FOR INTERACTIVE SEGMENTATION OF OBJECT ON IMAGE AND ELECTRONIC COMPUTING DEVICE FOR REALIZING SAID OBJECT | 2020 |
|
RU2742701C1 |
| SEGMENTATION AND PREDICTION OF TIME PATTERNS OF LOW LEVEL TRAILS | 2020 |
|
RU2783420C1 |
| METHOD FOR DETERMINING FAULT POSITION BASED ON SEISMIC DATA | 2022 |
|
RU2783367C1 |
| METHOD FOR DETERMINING DEFECTS IN THE IMAGE OF COMPOSITE PRODUCTS | 2022 |
|
RU2807288C1 |
| METHOD AND APPARATUS FOR IDENTIFYING PORTIONS OF FRAGMENTED MATERIAL WITHIN AN IMAGE | 2016 |
|
RU2694021C1 |
