[Paper Review] Scilab and SIP for Image Processing

Ricardo Fabbri, Odemir Martinez Bruno+1 more|arXiv (Cornell University)|Mar 18, 2012

Digital Image Processing Techniques7 references3 citations

TL;DR

This paper presents SIP (Scilab Image Processing Toolbox), a free, open-source extension to Scilab that enables efficient image processing and analysis through a high-level, MATLAB-like environment. It demonstrates how SIP simplifies complex tasks—such as edge detection, segmentation via watershed transformation, and multiscale skeletonization—by leveraging built-in functions for over 100 image formats and advanced algorithms, achieving robust results even in challenging cases like overlapping cell detection.

ABSTRACT

This paper is an overview of Image Processing and Analysis using Scilab, a free prototyping environment for numerical calculations similar to Matlab. We demonstrate the capabilities of SIP -- the Scilab Image Processing Toolbox -- which extends Scilab with many functions to read and write images in over 100 major file formats, including PNG, JPEG, BMP, and TIFF. It also provides routines for image filtering, edge detection, blurring, segmentation, shape analysis, and image recognition. Basic directions to install Scilab and SIP are given, and also a mini-tutorial on Scilab. Three practical examples of image analysis are presented, in increasing degrees of complexity, showing how advanced image analysis techniques seems uncomplicated in this environment.

Motivation & Objective

• To demonstrate the feasibility and advantages of using Scilab and its SIP toolbox for image processing and analysis in academic and industrial research.
• To address the complexity and high implementation burden of traditional image processing techniques in low-level languages like C or Fortran.
• To promote open-source software in image processing by showcasing SIP’s capabilities as a free, extensible, and user-friendly alternative to proprietary tools like MATLAB.
• To provide a practical, accessible environment for students and researchers to prototype and test advanced image analysis algorithms without high licensing costs.

Proposed method

• Utilizes Scilab’s interpreted, high-level programming language with MATLAB-like syntax for intuitive numerical and matrix-based image manipulation.
• Employs the SIP toolbox to read and write over 100 image formats via integration with the ImageMagick library.
• Applies standard image processing techniques including Gaussian blurring, median filtering, Canny and Sobel edge detection, and morphological operations.
• Uses the Euclidean distance transform and watershed segmentation to separate overlapping objects, with peak detection in the distance map used as markers.
• Implements a multi-step pipeline: image thresholding, inversion, distance transform with normalization, median filtering, and watershed segmentation with marker control.
• Visualizes results using 3D surface plots and random color mapping to distinguish segmented regions, enabling automatic object counting.

Experimental results

Research questions

• RQ1 Can a free, open-source environment like Scilab with SIP provide a viable, user-friendly alternative to proprietary image processing software such as MATLAB?
• RQ2 How effectively can SIP handle complex image analysis tasks such as segmentation of overlapping biological cells using standard algorithms like watershed transformation?
• RQ3 To what extent does the integration of external open-source libraries (e.g., ImageMagick, OpenCV) enhance SIP’s functionality and robustness?
• RQ4 How does the use of a high-level, interpreted language reduce development time and complexity in implementing advanced image processing algorithms?
• RQ5 What are the practical benefits of open-source software in scientific image processing, particularly in terms of extensibility, community support, and algorithm transparency?

Key findings

• SIP successfully supports over 100 major image file formats, including PNG, JPEG, BMP, and TIFF, through integration with ImageMagick, surpassing the format support of many proprietary tools.
• The watershed segmentation pipeline using distance transform and marker-controlled segmentation correctly separated overlapping blood cells in a real microscopy image, demonstrating robustness in complex cases.
• The method achieved accurate object counting via `n = maxi(w) - 1`, where `w` is the labeled watershed output, confirming the reliability of the segmentation process.
• The 3D visualization of the distance transform revealed clear peaks at the centers of each cell, validating the geometric basis of the segmentation approach.
• The use of a free, open-source stack (Scilab, SIP, ImageMagick, OpenCV) enabled full transparency, extensibility, and community-driven development without licensing or bureaucratic barriers.
• The toolbox has been adopted in universities and research settings worldwide—including in the USA, Germany, France, Greece, and Brazil—indicating strong real-world utility and acceptance.