题意：
使用python将一幅彩色图像转换为油画的效果图像
解析：
task1: 利用公式I(x,y) = 0.299∗r(x,y)+0.587∗g(x,y)+0.114∗b(x,y) 调整图像像素进行灰度化，对图像的行和列进行遍历，再分别对图像的r、g、b值进行换算，输出像素值rgb = np.uint8(r * 0.299 + b * 0.114 + g * 0.587)即可
task2： 目标是在邻域（包括当前像素）中找到最频繁的局部像素值以创建另一个图像J 计算每个像素的附近像素的强直方图对灰度等级进行统计 ，然后对模板进行遍历,使用直方图中高频像素替换J中对应窗口的中心像素
task3： 对于图像J(x,y)查找对应原始图像A的相应颜色强度，即J(x,y)=A(x,y)，计算每个波段的像素的三通道颜色平均强度去替换A ，得到的图像即为油画图像
涉及知识点：
python cv模块，图像像素处理
更多可加V讨论
V：tiamo-0620
COMP 9517
T3, 2019
Assignment 1: Specification
Maximum marks achievable: 10 marks
This assignment is worth 10% of the total course marks.
Deliverables: You will submit a report (maximum 5 pages) briefly explaining the approach you have taken in
Tasks 1, 2 and 3 and include some sample input images and the intermediate and final images obtained. You
must also submit the Python source code files.
Submission: The assignment files should be submitted online via WebCMS. Instructions for submission will be
posted closer to the deadline.
Tip: You are advised to use Opencv3 with python 3+. Jupyter notebook files are preferred as source
code files.
Objectives:
The objective of this assignment is to familiarise yourself with performing basic image processing operations
using OpenCV in Python. This assignment also introduces you to basic image analysis, and helps you gain
the following skills:
Opening and reading from image files
Displaying data as image files and writing image files
Performing simple mathematical operations on images
Applying image filters
Image manipulation and adjustment
The assignment files should be submitted online.
Instructions for submission will be posted closer to the deadline.
Deadline for submission is week 4 Monday Oct 7th, 23:59:59
Description:
Image processing techniques are used in digital art to represent digital images in more creative (and
alternative!) ways. The goal of this assignment is to open a digital image, perform a sequence of image
processing / manipulation operations (step-by-step as listed under Instructions) and produce a creative effect,
namely an oil painting-like output image.
Below is an original image, and the final `oil painting’ result obtained after applying a sequence of image
manipulation operations.
Instructions:
Task 1 (3 marks)
Two colour images (dog.jpg, light_rail.jpg) have been provided to you as sample inputs. You may also test
on other input images of your choice.
Starting with an input colour image (let us call this image A), you should combine the three colour-bands into
one band using the following equation:
I(x,y) = 0.299∗r(x,y)+0.587∗g(x,y)+0.114∗b(x,y)
Where r, g and b are colour-bands of image I.
You will go through the image pixel by pixel and perform this operation on each pixel. The resulting image
I will contain only one combined band, i.e. it is a grey-level image.
Retain a copy of the resultant image I.
Include I as an intermediate result in the report.
Task 2 (4 marks)
Once you obtain the combined single band image I, the next task is to perform the following sequence of
operations on I. The goal is to find the most frequent local pixel value in a neighbourhood (including the
current pixel) to create another image J.
To create image J, first make a copy of image I called J.
, you will:
first define an appropriate neighbourhood (window) around a pixel in I
compute the intensity histogram for pixels in the neighbourhood around each pixel-
pick the most frequent pixel value in the neighbourhood using the histogram, and replace the
central pixel in the corresponding window in J with the most frequent local pixel value for that
window
repeat the above for every pixel in I
Try at least 3 different window sizes in Task 2.
The resulting image J for each window size is the output for Task 2, which should be included in the report.
Any additional and meaningful intermediate results may also be included in the report.
Task 3 (3 marks)
Using J from Task 2, you will construct the final ‘oil painting effect’ image. To do this, you will
make a copy of the original image A (and let us call it image B)
for each pixel (x, y) in the image J, in the local neighbourhood of pixel (x, y) defined in Task 2
• find all pixels in J which have the same value of J (x, y)
• get the corresponding colour intensities of those pixels in image B
• calculate the average intensities of those pixels in each band
• then replace B (x, y) with the average intensities in each corresponding band
Try the same window sizes here as in Task 2, to experiment with the art effects produced. The resulting
image B for different window sizes is the output for Task 3 and should be included in the report.
Compulsory output images to be included in the report (for each window size): intermediate outputs I
and J, oil painting-like image B.
© Copyright: Arcot Sowmya, CSE, UNSW, with acknowledgements to COMP9517
teaching team past and present.
20 Sept 2019
