Histogram—Representation of Color Feature in Image Processing Yang, Li

Slides:



Advertisements
Similar presentations
Distinctive Image Features from Scale-Invariant Keypoints David Lowe.
Advertisements

13- 1 Chapter 13: Color Processing 。 Color: An important descriptor of the world 。 The world is itself colorless 。 Color is caused by the vision system.
Presented by Xinyu Chang
November 12, 2013Computer Vision Lecture 12: Texture 1Signature Another popular method of representing shape is called the signature. In order to compute.
Wen-Hung Liao Department of Computer Science National Chengchi University November 27, 2008 Estimation of Skin Color Range Using Achromatic Features.
Hongliang Li, Senior Member, IEEE, Linfeng Xu, Member, IEEE, and Guanghui Liu Face Hallucination via Similarity Constraints.
Image Processing IB Paper 8 – Part A Ognjen Arandjelović Ognjen Arandjelović
COLORCOLOR A SET OF CODES GENERATED BY THE BRAİN How do you quantify? How do you use?
Object Recognition & Model Based Tracking © Danica Kragic Tracking system.
Color Image Processing
Aalborg University Copenhagen
Integrating Color And Spatial Information for CBIR NTUT CSIE D.W. Lin
SWE 423: Multimedia Systems Chapter 4: Graphics and Images (2)
A Study of Approaches for Object Recognition
Color.
Color: Readings: Ch 6: color spaces color histograms color segmentation.
CS292 Computational Vision and Language Visual Features - Colour and Texture.
CSE 803 Stockman Fall Color Used heavily in human vision Color is a pixel property, making some recognition problems easy Visible spectrum for humans.
Images and colour Colour - colours - colour spaces - colour models Raster data - image representations - single and multi-band (multi-channel) images -
Real-Time Face Detection and Tracking Using Multiple Cameras RIT Computer Engineering Senior Design Project John RuppertJustin HnatowJared Holsopple This.
CS559-Computer Graphics Copyright Stephen Chenney Color Recap The physical description of color is as a spectrum: the intensity of light at each wavelength.
Image Processing David Kauchak cs458 Fall 2012 Empirical Evaluation of Dissimilarity Measures for Color and Texture Jan Puzicha, Joachim M. Buhmann, Yossi.
9/14/04© University of Wisconsin, CS559 Spring 2004 Last Time Intensity perception – the importance of ratios Dynamic Range – what it means and some of.
7.3 การประมวลผล ภาพสี Color Image Processing. Color is a perceptual manifestation of light which in turn is an electromagnetic signal. Color is a sensation.
Multimedia and Time-series Data
I-1 Steps of Image Generation –Create a model of the objects –Create a model for the illumination of the objects –Create an image (render) the result I.
1 © 2010 Cengage Learning Engineering. All Rights Reserved. 1 Introduction to Digital Image Processing with MATLAB ® Asia Edition McAndrew ‧ Wang ‧ Tseng.
Image Retrieval Part I (Introduction). 2 Image Understanding Functions Image indexing similarity matching image retrieval (content-based method)
Content-Based Image Retrieval
Texture. Texture is an innate property of all surfaces (clouds, trees, bricks, hair etc…). It refers to visual patterns of homogeneity and does not result.
COLOR HISTOGRAM AND DISCRETE COSINE TRANSFORM FOR COLOR IMAGE RETRIEVAL Presented by 2006/8.
Color Theory ‣ What is color? ‣ How do we perceive it? ‣ How do we describe and match colors? ‣ Color spaces.
September 17, 2013Computer Vision Lecture 5: Image Filtering 1ColorRGB HSI.
Advanced Multimedia Image Content Analysis Tamara Berg.
Graphics Lecture 4: Slide 1 Interactive Computer Graphics Lecture 4: Colour.
1 Similarity-based matching for face authentication Christophe Rosenberger Luc Brun ICPR 2008.
Gang WangDerek HoiemDavid Forsyth. INTRODUCTION APROACH (implement detail) EXPERIMENTS CONCLUSION.
November 30, PATTERN RECOGNITION. November 30, TEXTURE CLASSIFICATION PROJECT Characterize each texture so as to differentiate it from one.
Digital Image Processing In The Name Of God Digital Image Processing Lecture6: Color Image Processing M. Ghelich Oghli By: M. Ghelich Oghli
Introduction to Computer Graphics
June 14, ‘99 COLORS IN MATLAB.
Timo Ahonen, Abdenour Hadid, and Matti Pietikainen
A Tutorial on using SIFT Presented by Jimmy Huff (Slightly modified by Josiah Yoder for Winter )
Colour and Texture. Extract 3-D information Using Vision Extract 3-D information for performing certain tasks such as manipulation, navigation, and recognition.
Intelligent Robotics Today: Vision & Time & Space Complexity.
Image features and properties. Image content representation The simplest representation of an image pattern is to list image pixels, one after the other.
Cell Segmentation in Microscopy Imagery Using a Bag of Local Bayesian Classifiers Zhaozheng Yin RI/CMU, Fall 2009.
Learning and Removing Cast Shadows through a Multidistribution Approach Nicolas Martel-Brisson, Andre Zaccarin IEEE TRANSACTIONS ON PATTERN ANALYSIS AND.
Color Models Light property Color models.
Half Toning Dithering RGB CMYK Models
Color Image Processing
Color Image Processing
Images In Matlab.
Chapter 6: Color Image Processing
Color Image Processing
PRESENTED BY Yang Jiao Timo Ahonen, Matti Pietikainen
Colour Theory Fundamentals
CITA 342 Section 7 Working with Color.
Color: Readings: Ch 6: color spaces color histograms
Color Representation Although we can differentiate a hundred different grey-levels, we can easily differentiate thousands of colors.
Computer Vision Lecture 4: Color
Color: Readings: Ch 6: color spaces color histograms
Fall 2012 Longin Jan Latecki
Color Image Processing
Slides taken from Scott Schaefer
The SIFT (Scale Invariant Feature Transform) Detector and Descriptor
Estimation of Skin Color Range Using Achromatic Features
Color Image Processing
Digital Image Processing Lecture 26: Color Processing
Color Model By : Mustafa Salam.
Presentation transcript:

Histogram—Representation of Color Feature in Image Processing Yang, Li Part II: Histogram—Representation of Color Feature in Image Processing Yang, Li Good afternoon, the research interest in this part focus on the histogram, the representation of color feature in image processing. Let have a look at the recent development in this field.

Structure: Color histogram descriptor Color cooccurrence histogram HSV space segmentation

Color histogram descriptor <C, M,{Ci}, {H(Ci)}> A weighted Euclidean distance of colors is then taken to match color histograms, if X is the query histogram and Y is the histogram of an item in the database, then the similarity between X and Y is given by:||Z||=Z'AZ Searching and locating multimedia data need a good description and representation of multimedia information. Mahood and Tanveer propose two ways of capturing color content in image called Color Histogram “and Region Color.” The descriptors are suitable for a wide variety of applications requiring image-to-image matching and object-to-image matching. *Color histogram is a record of the number of image or region pixels of a specified color. It can be described as a tuple. C is chosen color space, Ci is the color quantization cell, M is the number of of color. H(Ci) = Ni/N, and Ni is the number of image pixels whose color falls in the bin Ci and N is the total number of image pixels. A is a symmetric color similarity matrix in which a(i,j) indicates the similarity of colors i and j in the histogram. This method accounts for both the perceptual distance between different pairs of colors(e.g. orange and red are less different than orange and blue), and the difference in the amounts of a given color(e.g. particular shade of red).

Region-Color Descriptor The region information in both query and image can be represented using the region color descriptor with the difference that the number of regions and their respective colors are different. For a set of image regions though, the color label or indes assigned to the corresponding regions must be the same. When the goal is to find a query embedded in an image based on the color of one or more of its regions, we need a robust description of a region color that is illumination and pose-invariant to account for the different appearances of the query object in images of the database. Ri is a region in the image and Cj is its color description. The region information in both query and image can be represented using the region color descriptor with the difference that the number of regions and their respective colors are different. For a set of query color regions to be accurately detected in the set of image regions though, the color label or index assigned to the corresponding regions must be the same. In other words, any method of describing the region’s color must be illumination and pose-invariant to account for the various appearances of the query object under different illumination conditions.

Color Cooccurrence Histogram Each model image is represented as a color CH. The color CH holds the number of occurrences of pairs of color pixels C1=(R1,G1,B1) and C2=(R2,G2,B2) separated by a vector in the image y plane(x, y). The color cooccurrence histogram keeps track of the number of pairs of certain colored pixels that occur at certain separation distances in image space. The color cooccurrence histogram adds geometric information to the normal color histogram. The model cooccurrence histogram are matched to subregions in test images to find the object. By adjusting the number of colors and the number of distances used in the CH, the tolerance of the algorithm to the changes in lighting, viewpoint, and the flexibility of the object can be adjusted. C2 y C1 x

Quantize colors into a set of representative colors C=(c1,c2,…,c ) Assumption: ignore the direction of(x, y) and keep track of only magnitude d= Quantize colors into a set of representative colors C=(c1,c2,…,c ) Quantize the distances into a set of distance ranges D={[0,1),[1,2),…,[ -1, )}. CH is represented by CH (i, j, k). Where i and j index the two colors from set C, and k indexes the distance range from set D. In searching an image for an object, the image is scanned for a rectangular window that gives a CH similar to one of the training CHs.

It indicates how well the image CH accounts for the model CH. The image and model CHs are compared by computing their intersection. The intersection is: It indicates how well the image CH accounts for the model CH. If the image accounts for all the entries in the model CH, then the intersection will be equal to the sum of all the entries in in the model CH, or Imm. If the

The conversion form RGB to HSV is performed with the equations: Recursive HSV-space segmentation to extract regions within the image which contain perceptually similar color The conversion form RGB to HSV is performed with the equations: Where H=H1 if B<=G; otherwise H=360-H1; Where R,G,B are the red, green, and blue component values which exist in the range [0,255]. H: hue S:saturate V:value

If the colors with value<25%, they are classified as black; if the color with saturation<20% and value>75%, they can be classified as white. VAL white Green 120 Yellow 60 Blue 240 Magenta 300 BRIGHT CHROMATIC CHROMATIC black hue SAT

White Black chromatic yes yes no Value<25 IMAGE no no bright SAT<20 Value<25 IMAGE no no Value>75 SAT>=20 chromatic yes bright chromatic Build Hue Histogram Build Saturation Histogram Determine N peaks Determine M peaks Each image is examined to classify the pixels into one of the four categories: Black, White, Bright chromatic, and Chromatic regions. If the If the colors with value<25%, they are classified as black; if the color with saturation<20% and value>75%, they can be classified as white. All the remaining pixels fall in the chromatic region of the HSV cone, hue and saturation histogram are built and the corresponding peaks are threshhold to segment the colors. Threshhold Peak i Threshhold Peak j no no i=n? i=m ? yes yes END