Scanning Kernel

Lit review

• Object Categorization by Learned Universal Visual Dictionary
  • 2005 Tenth IEEE International Conference on Computer Vision (ICCV'05)
  • This paper presents a new algorithm for the automatic recognition of object classes from images (categorization). Compact and yet discriminative appearance-based object class models are automatically learned from a set of training images. The method is simple and extremely fast, making it suitable for many applications such as semantic image retrieval, web search, and interactive image editing. It classifies a region according to the proportions of different visual words (clusters in feature space). The specific visual words and the typical proportions in each object are learned from a segmented training set. The main contribution of this paper is two fold: i) an optimally compact visual dictionary is learned by pair-wise merging of visual words from an initially large dictionary. The final visual words are described by GMMs. ii) A novel statistical measure of discrimination is proposed which is optimized by each merge operation. High classification accuracy is demonstrated for nine object classes on photographs of real objects viewed under general lighting conditions, poses and viewpoints. The set of test images used for validation comprise: i) photographs acquired by us, ii) images from the web and iii) images from the recently released Pascal dataset. The proposed algorithm performs well on both texture-rich objects (e.g. grass, sky, trees) and structure-rich ones (e.g. cars, bikes, planes).

Existing Software

• CellProfiler
• ITK

Terms

• Image cytometry = automated cell image analysis
• High throughput
• fluorescently labelled cellular components
• Segmentation is the process of identifying and classifying data found in a digi- tally sampled representation. Typically the sampled representation is an image acquired from such medical instrumentation as CT or MRI scanners.

Concepts

• Quantitative image assays part of
• Concept of streaming = the ability to automatically break data into smaller pieces, process the pieces one by one, and reassemble the processed data into a final result.
• Erosion and dilation
• Voting filter for noise reduction
• Variety of image modalities

General (ITK-implemented) Segmentation Techniques

• Region Growing
  • Connected threshold - evaluation of pixel intensity
  • Otsu segmentation
• Watershed
• Level Set
• Shape Detection

Abstract

This poster presents a new software tool based on a novel algorithm for the automatic recognition and segmentation of 2D-image data that is suitable for any image modality in a wide variety of applications. It uses an expert-trained visual dictionary of morphology, generated simply by drawing bounding boxes around morphology representing an object class of interest. The tool then systematically crawls an input image performing classifications on the local region using the highly accurate WND-CHARM algorithm.

If a biologist wants to generate an assay for a tissue sample, he'll have to create a specific antibody fluorescent probe that binds to the protein of interest. Based on an expert trained visual dictionary, simply by drawing a bounding box around a region that belongs in a training set. And using machine learning

Simply defining a dictionary of images that reasonably cover the range of morphological possibilities, Crawl a window over a larger... neighborhood, process image region.

Here we present a novel tool kernel classifier for biologists to probe larger images for items of interest, simply by training a classifier. Segmentation without the need for seed points.

A tool for biologists to to systematically probe large 2-D images Image classification using training image sets