I'm just trying to adjust contrast/ brightness in an image in gray scale to highlight whites in that image with Opencv in C. How can I do that? is there any function that makes this task in opencv?
Original image:
Modified image:
Thanks in advance!
I think you can adjust contrast here in two ways:
1) Histogram Equalization :
But when i tried this with your image, result was not as you expected. Check it below:
2) Thresholding :
Here, i compared each pixel value of input with an arbitrary value ( which i took 127). Below is the logic which has inbuilt function in opencv. But remember, output is Binary image, not grayscale as you did.
If (input pixel value >= 127):
ouput pixel value = 255
else:
output pixel value = 0
And below is the result i got :
For this, you can use Threshold function or compare function
3) If you are compulsory to get grayscale image as output, do as follows:
(code is in OpenCV-Python, but for every-function, corresponding C functions are available in opencv.itseez.com)
for each pixel in image:
if pixel value >= 127: add 'x' to pixel value.
else : subtract 'x' from pixel value.
( 'x' is an arbitrary value.) Thus difference between light and dark pixels increases.
img = cv2.imread('brain.jpg',0)
bigmask = cv2.compare(img,np.uint8([127]),cv2.CMP_GE)
smallmask = cv2.bitwise_not(bigmask)
x = np.uint8([90])
big = cv2.add(img,x,mask = bigmask)
small = cv2.subtract(img,x,mask = smallmask)
res = cv2.add(big,small)
And below is the result obtained:
You could also check out the OpenCV CLAHE algorithm. Instead of equalizing the histogram globally, it splits up the image into tiles and equalizes those locally, then stitches them together. This can give a much better result.
With your image in OpenCV 3.0.0:
import cv2
inp = cv2.imread('inp.jpg',0)
clahe = cv2.createCLAHE(clipLimit=4.0, tileGridSize=(8,8))
res = clahe.apply(inp)
cv2.imwrite('res.jpg', res)
Gives something pretty nice
Read more about it here, though it's not super helpful:
http://docs.opencv.org/3.1.0/d5/daf/tutorial_py_histogram_equalization.html#gsc.tab=0
Although this post is a bit aged:
What about using "cvAddWeighted( )" ?
What it does is:
dst = src1*alpha + src2*beta + gamma
What I understand from applying brightness and contrast is, that one wants to do:
dst = src*contrast + brightness;
so if
src1 = input image
src2 = any image of same type as src1
alpha = contrast value
beta = 0.0
gamma = brightness value
dst = resulting Image (must be of same type as src1)
One should be pretty much done with the task, no?
This approch works for me using CvMat* images
Related
I was experimenting with GameplayKit’s GKAgent3D class to move a SCNNode within a scene. I was able to update the SCNNode with the agent’s position, but not rotation. The issue being the agent’s rotation is stored as a matrix_float3x3, which doesn’t match any of data types SceneKit uses for storing rotational information.
So what I’d like to know is if there’s a simple function or method that could convert a rotation stored as matrix_float3x3 to any SceneKit data types?
To expand on #rickster 's answer, here's a nice way to take the top-left 3x3 of a 4x4 matrix in Swift, taking advantage of the expanded simd support in the iOS 11/ tvOS 11/ High Sierra version of SceneKit:
extension float4 {
var xyz: float3 {
return float3(x, y, z)
}
init(_ vec3: float3, _ w: Float) {
self = float4(vec3.x, vec3.y, vec3.z, w)
}
}
extension float4x4 {
var upperLeft3x3: float3x3 {
let (a,b,c,_) = columns
return float3x3(a.xyz, b.xyz, c.xyz)
}
init(rotation: float3x3, position: float3) {
let (a,b,c) = rotation.columns
self = float4x4(float4(a, 0),
float4(b, 0),
float4(c, 0),
float4(position, 1))
}
}
Then, to update your agent to match your node's orientation, you'd write:
agent.rotation = node.simdTransform.upperLeft3x3
Or, if the node in question is not at the "root" level (as in, a direct child of the rootNode), you might want to use the node's worldTransform:
agent.rotation = node.simdWorldTransform.upperLeft3x3
EDIT: If the node in question has a dynamic physics body attached, or is being animated with an SCNTransaction block, the node's presentation node will more accurately reflect its current position on screen:
agent.position = node.presentation.simdWorldPosition
agent.rotation = node.presentation.simdWorldTransform.upperLeft3x3
EDIT: added code above for going in the other direction, moving the node to match the agent.
node.simdTransform = float4x4(rotation: agent3d.rotation, position: agent3d.position)
Note that if you have a physics body attached to the node, it should be kinematic rather than dynamic if you're going to be directly modifying the node's transform in this way.
SceneKit takes transform matrices as SCNMatrix4, and provides utilities for converting from SIMD matrix_float4x4: init(_ m: float4x4) for Swift and SCNMatrix4FromMat4 for ObjC/C++.
Sadly, I don't see a built-in way to convert between SIMD 3x3 and 4x4 matrices using the assumption that the 3x3 is the upper left of the 4x4. (Seems like you'd expect that in the SIMD library, so it's worth filing a bug to Apple about.)
But it's not too hard to provide one yourself: just construct a 4x4 from column vectors, using the three column vectors of the 3x3 (padded out to float4 vectors with zero for the w component) and identity for the fourth column (0,0,0,1). (Implementation code left for the reader, partly because I don't want to write it for three languages.) After converting float3x3 to float4x4 you can convert to SCNMatrix4.
Edit: In iOS 11 / tvOS 11 / macOS 10.13 (why didn't they just call this year's macOS version 11, too?), SceneKit has a whole parallel set of APIs for using SIMD types like float4x4 directly; e.g. simdTransform. However, you still need to convert a 3x3 to a 4x4 matrix.
first of all, I'm totally new to kivy, so I'm struggling a bit.
I'm trying to display a numpy array in a kivy window.
So far i figured out that this should work using the Texture Class (http://kivy.org/docs/api-kivy.graphics.texture.html).
As my numpy array changes from time to time, I'm trying to adjust the following code to my application.
# create a 64x64 texture, defaults to rgb / ubyte
texture = Texture.create(size=(64, 64))
# create 64x64 rgb tab, and fill with values from 0 to 255
# we'll have a gradient from black to white
size = 64 * 64 * 3
buf = [int(x * 255 / size) for x in range(size)]
# then, convert the array to a ubyte string
buf = b''.join(map(chr, buf))
# then blit the buffer
texture.blit_buffer(buf, colorfmt='rgb', bufferfmt='ubyte')
# that's all ! you can use it in your graphics now :)
# if self is a widget, you can do this
with self.canvas:
Rectangle(texture=texture, pos=self.pos, size=(64, 64))
It seems that creating the texture and changing it works as it should, but i dont get, how to display the texture.
Can anybody explain to me, how to use the
with self.canvas:
Rectangle(texture=texture, pos=self.pos, size=(64, 64))
in a way, that I get to see my picture/numpy array.
Thanks alot in advance!
Holzroller
Edit:
I figured out that using Kivy 1.8.0 and the Texture Class is a bit messy. So I upgraded to Kivy 1.9.0 via github (installing Kivy via apt-get in Ubuntu 14.04 LTS serves you the 1.8.0 version) and I get to see the Texture using the following code. I hope that helps people who are having the same problem as me.
from kivy.graphics.texture import Texture
from kivy.graphics import Rectangle
from kivy.uix.widget import Widget
from kivy.base import runTouchApp
from array import array
from kivy.core.window import Window
# create a 64x64 texture, defaults to rgb / ubyte
texture = Texture.create(size=(1280, 1024), colorfmt='rgb')
# create 64x64 rgb tab, and fill with values from 0 to 255
# we'll have a gradient from black to white
size = 1280 * 1024 * 3
buf = [int(x * 255 / size) for x in range(size)]
# then, convert the array to a ubyte string
arr = array('B', buf)
# buf = b''.join(map(chr, buf))
# then blit the buffer
texture.blit_buffer(arr, colorfmt='rgb', bufferfmt='ubyte')
# that's all ! you can use it in your graphics now :)
# if self is a widget, you can do this
root = Widget()
with root.canvas:
Rectangle(texture=texture, pos=(0, 0), size=(1280*3, 1024*3))
runTouchApp(root)
Edit2:
Basically I'm back to the original Problem:
I have a numpy array (type 'numpy.ndarray'; dtype 'uint8') and I'm trying to convert it into a format, so that the texture will show me the image. I tried to break it down to the same way it is done in the example code i posted above. But i sadly doesn't work. I really do not know what I'm doing wrong here.
(my numpy array is called im2 in the folling code)
list1 = numpy.array(im2).reshape(-1,).tolist()
arr = array('B', list1)
texture.blit_buffer(arr, colorfmt='rgb', bufferfmt='ubyte')
Numpy have a tostring() attribute, that you could use directly, if the source array is uint8 type. You don't even need to reshape:
texture = Texture.create(size=(16, 16), colorfmt="rgb"))
arr = numpy.ndarray(shape=[16, 16, 3], dtype=numpy.uint8)
# fill your numpy array here
data = arr.tostring()
texture.blit_buffer(data, bufferfmt="ubyte", colorfmt="rgb"
About the issue you're talking in the comment, i see 2 points:
Ensure the callback from the ROS is called in the mainthread. Maybe the update is simply ignored.
When you manually change inplace the texture, the associated object that use it are not notified, you need to do it. Add a self.canvas.ask_update() to ensure the canvas redisplay at the next frame.
I want to re-project an HDF from UTM(WGS84) to sinusoidal（WGS84), so I try to use GDALAutoCreateWarpedVRT to finished it. The code is below:
hSrcDS = (GDALDataset*)GDALOpen("HJ1ACCD1.hdf", GA_ReadOnly);
const char *pszSrcWKT = NULL;
char* pszDstWKT = NULL;
//pszSrcWKT = ProjectionStr;
pszSrcWKT=GDALGetProjectionRef(hSrcDS);
CPLAssert( pszSrcWKT != NULL &&strlen(pszSrcWKT) > 0 );
OGRSpatialReference oSRS;
oSRS.SetSinusoidal(0,0,0);
oSRS.SetWellKnownGeogCS("WGS84");
oSRS.exportToWkt(&pszDstWKT );
GDALWarpOptions*psWarpOptions = GDALCreateWarpOptions();
psWarpOptions->dfWarpMemoryLimit=500*1024*1024;
hDstDS=(GDALDataset*)(GDALDataset*)GDALAutoCreateWarpedVRT(hSrcDS,pszSrcWKT,pszDstWKT,GRA_Bilinear ,20,psWarpOptions);
GDALDriver *poDriverTiff;
poDriverTiff=GetGDALDriverManager()->GetDriverByName("GTIFF");
poDriverTiff->CreateCopy("d:\\toto.tif",(GDALDataset*)hDstDS,false,NULL,NULL,NULL);
When I set oSRS.SetSinusoidal(0,0,0)，the result seems good, but the resolution is doubled (from 30 to 60). It's so weird.
According to the API docs for GDALAutoCreateWarpedVRT:
The GDALSuggestedWarpOutput() function is used to determine the bounds and resolution of the output virtual file which should be large enough to include all the input image
There is also a GDALSuggestedWarpOutput2() function to help suggest output file size for a similar set of requirements.
I am trying to write a program that will recognize an image on the screen, compare it against a resource library, and then calculate based on the result of the image source.
The first thing that I did was to create the capture screen function which looks like this:
private Bitmap Screenshot()
{
System.Drawing.Bitmap Table = new System.Drawing.Bitmap(88, 40, PixelFormat.Format32bppArgb);
System.Drawing.Graphics g = System.Drawing.Graphics.FromImage(RouletteTable);
g.CopyFromScreen(1047, 44, 0, 0, Screen.PrimaryScreen.Bounds.Size);
return Table;
}
Then, I analyze this picture. The first method I used was to create two for loops and analyze both the bitmaps pixel by pixel. The problem with this method was time, it took a long time to complete 37 times. I looked around and found the convert to bytes and the convert to hash methods. This is the result:
public enum CompareResult
{
ciCompareOk,
ciPixelMismatch,
ciSizeMismatch
};
public CompareResult Compare(Bitmap bmp1, Bitmap bmp2)
{
CompareResult cr = CompareResult.ciCompareOk;
//Test to see if we have the same size of image
if (bmp1.Size != bmp2.Size)
{
cr = CompareResult.ciSizeMismatch;
}
else
{
//Convert each image to a byte array
System.Drawing.ImageConverter ic = new System.Drawing.ImageConverter();
byte[] btImage1 = new byte[1];
btImage1 = (byte[])ic.ConvertTo(bmp1, btImage1.GetType());
byte[] btImage2 = new byte[1];
btImage2 = (byte[])ic.ConvertTo(bmp2, btImage2.GetType());
//Compute a hash for each image
SHA256Managed shaM = new SHA256Managed();
byte[] hash1 = shaM.ComputeHash(btImage1);
byte[] hash2 = shaM.ComputeHash(btImage2);
for (int i = 0; i < hash1.Length && i < hash2.Length&& cr == CompareResult.ciCompareOk; i++)
{
if (hash1[i] != hash2[i])
cr = CompareResult.ciPixelMismatch;
}
}
return cr;
}
After I analyze the two bitmaps in this function, I call it in my main form with the following:
Bitmap Table = Screenshot();
CompareResult success0 = Compare(Properties.Resources.Result0, Table);
if (success0 == CompareResult.ciCompareOk)
{ double result = 0; Num.Text = result.ToString(); goto end; }
The problem I am getting is that once this has all been accomplished, I am always getting a cr value of ciPixelMismatch. I cannot get the images to match, even though the images are identical.
To give you a bit more background on the two bitmaps, they are approximately 88 by 40 pixels, and located at 1047, 44 on the screen. I wrote a part of the program to automatically take a picture of that area so I did not have to worry about the wrong location or size being captured:
Table.Save("table.bmp");
After I took the picture and saved it, I moved it from the bin folder in the project directly to the resource folder and ran the program again. Despite all of this, the result is still ciPixelMismatch. I believe the problem lies within the format that the pictures are being saved as. I believe that despite them being the same image, they are being analyzed in different formats, maybe one of the pictures contains a bit more information than the other which is causing the mismatch. Can somebody please help me solve this problem? I am just beginning with my c# programming, I am 5 days into the learning process, and I am really at a loss for this.
Yours sincerely,
Samuel
I'm trying to convert the byteArray of a Sound Object to an array with floats. The Sound Object plays back fine & at full length, but the float Array i get from it is cut off (but sounds correct), so i must be doing something wrong in the conversion:
var s:Sound = mySound;
s.play(); // plays fine
var bytes:ByteArray = new ByteArray();
bytes.endian = Endian.LITTLE_ENDIAN;
s.extract(bytes, s.bytesTotal, 0);
var leftChannel:Array = new Array();
var rightChannel:Array = new Array();
bytes.position = 0;
while (bytes.bytesAvailable)
{
leftChannel.push(bytes.readFloat());
rightChannel.push(bytes.readFloat());
}
and this is what i get:
The top two channels are the original Sound Object.
The lower two is the float Array Data. I aligned them so you can see that the beginning is cut off and obviously the length is incorrect.
Thanks for any answers...
ok there were two problems:
the mp3 file i was importing was somehow corrupt, that caused the beginning to be cut off
the length i defined to extract was not correct, to find the full sound length use
var numTotalSamples:Number = int(s.length * 44.1); //assuming 44.1kHz sample rate
then:
s.extract(bytes, numTotalSamples, 0);