CelexMatlabToolbox

中文README

Introduction

CelexMatlabToolbox is a matlab toolbox to process the events collected using Celex IV Dynamic Vision Sensor.

File structure and Functions

• createImgFromRawData.m
  • Real-time decoding of events from raw data.
  • Real-time showing and saving binary pics from raw data.
  • Real-time showing and saving gray pics from raw data.
  • Real-time showing and saving accumulated gray pics from raw data.
• functions: a collection of matlab functions.
  • Decode from raw data(.bin file) in batches and return events in format of x,y,adc,t，where t is the continuous time stamp.
  • Save decoded events as mat file in memory-efficient way.
  • Load events from mat file.
  • Show and save binary pics from events.
  • Show and save gray pics from events.
  • Show and save accumulated gray pics from events.
  • Show and save denoised binary pics from events.
  • Show and save denoised gray pics from events.
  • Display 3D events flow from events.
  • Accumulate frames by specified time interval in certain range (in percentage), and user can skip the first specified amount of frames if needed and also determine the time for each frame to show.
• demo.m
  • An executable demo file, which includes all sample usage to call the functions in folder functions.

Sample Results

• Denoising

• 3D Events flow

Usage

• Before you use
  • Right click on the CelexMatlabToolbox folder to include folder and its subfolders.
  • Modify the directory and filepath in demo.m .
• For each function in functions folder, the description of its corresponding function, input and output is offered. To read the description, please run help <functionName> ，for instancehelp showAllPic.
• You can find sample usage of all functions in demo.m.
  • You will find many code sections that are seperated by %% in file demo.py, choose the code section you want to run, then click the Run code section button in the toolbar to execute the functions of that section.

TODO

• Visualizing by accumulating events through a specified time interval.
• Take spatial connection into consideration to improve denoising results.
• Take FPN into consideration to improve denoising results.
• Add overlap parameter for visualizing.

LICENSE

• Apache-2.0