Deteksi Kematangan Buah Jambu Kristal Berdasarkan Fitur Warna Menggunakan Metode Transformasi Ruang Warna Hsv (Hue Saturation Value) Dan K-Nearest Neighbor

Guava has properties that are easily damaged, improper handling of guava fruit can result in a decrease in quality and quality. In general, to measure maturity is still done manually, the weakness of this method is the level of accuracy that is not consistent and tends to experience errors. Utilization of images is very important to determine the maturity of guava fruit by utilizing digital images. With the existence of digital images, to determine the maturity of guava fruit based on its color, it can be done computing (technology-based), namely by applying image processing using the HSV (Hue, Saturation, Value) color space transformation method. The HSV (Hue, Saturation, Value) color model groups the intensity components of the carried color information (hue and saturation) in image colors. The results of the ripeness detection can be seen in each test with a percentage value of 91.67% for the ripe guava category, 90% for the raw guava fruit category. The percentage value for testing the overall data has a good percentage value which is influential in detecting the maturity of crystal guava, which is 95%. So it can be concluded that the detection of ripeness of crystal guava fruit can be done by applying the HSV color space transformation method.

CMYK-CIELAB Color Space Transformation Using Machine Learning Techniques

In this work four different machine learning approaches have been implemented to perform the color space transformation between CMYK and CIELAB color spaces. We have explored the performance of Support-Vector Regression (SVR), Artificial Neural Networks (ANN), Deep Neural Networks (DNN), and Radial Basis Function (RBF) models to achieve this color space transformation, both AToB and BToA direction. The data set used for this work was FOGRA53 which is composed of 1617 color samples represented both in CMYK and CIELAB color space values. The accuracy of the transformation models was measured in terms of ΔE* color difference. Moreover, the proposed models were compared, in practical terms, with the performance of the standard ICC profile for this color space transformation. The results showed that, for the forward transformation (CMYK to CIELAB), the highest accuracy was obtained using RBF. While, for the backward transformation (CIELAB to CMYK), the highest accuracy was obtained with DNN.

A Correspondence-Free Color Chart Design for Color Calibration

Colorimetric calibration computes the necessary color space transformation to map a camera's device-specific color space to a device-independent perceptual color space. Color calibration is most commonly performed by imaging a color rendition chart with a fixed number of color patches with known colorimetric values (e. g., CIE XYZ values). The color space transformation is estimated based on the correspondences between the camera's image and the chart's colors. We present a new approach to colorimetric calibration that does not require explicit color correspondences. Our approach computes a color space transformation by aligning the color distributions of the captured image to the known distribution of a calibration chart containing thousands of colors. We show that a histogram-based colorimetric calibration approach provides results that are onpar with the traditional patch-based method without the need to establish correspondences.

Color Edge Detection Based on the Fusion of Intensity and Chromatic Differences

Edge detection is the name for a set of mathematical methods which target at classifying points in an image at which the image intensity varies sharply or, has discontinuities. The paper tries to find the solution for detecting color edges based on color and intensity information of two new images H-image and T-image crafted on color space transformation, that will produce two-resulted edges derivates of H-image and T-image and are at last coalesced to obtain final edge.

Reversible denoising and lifting based color component transformation for lossless image compression

An undesirable side effect of reversible color space transformation, which consists of lifting steps (LSs), is that while removing correlation it contaminates transformed components with noise from other components. Noise affects particularly adversely the compression ratios of lossless compression algorithms. To remove correlation without increasing noise, a reversible denoising and lifting step (RDLS) was proposed that integrates denoising filters into LS. Applying RDLS to color space transformation results in a new image component transformation that is perfectly reversible despite involving the inherently irreversible denoising; the first application of such a transformation is presented in this paper. For the JPEG-LS, JPEG 2000, and JPEG XR standard algorithms in lossless mode, the application of RDLS to the RDgDb color space transformation with simple denoising filters is especially effective for images in the native optical resolution of acquisition devices. It results in improving compression ratios of all those images in cases when unmodified color space transformation either improves or worsens ratios compared with the untransformed image. The average improvement is 5.0–6.0% for two out of the three sets of such images, whereas average ratios of images from standard test-sets are improved by up to 2.2%. For the efficient image-adaptive determination of filters for RDLS, a couple of fast entropy-based estimators of compression effects that may be used independently of the actual compression algorithm are investigated and an immediate filter selection method based on the detector precision characteristic model driven by image acquisition parameters is introduced.

Color Space Transformation using Neural Networks

We investigated how well a multilayer neural network could implement the mapping between two trichromatic color spaces, specifically from camera R,G,B to tristimulus X,Y,Z. For training the network, a set of 800,000 synthetic reflectance spectra was generated. For testing the network, a set of 8,714 real reflectance spectra was collated from instrumental measurements on textiles, paints and natural materials. Various network architectures were tested, with both linear and sigmoidal activations. Results show that over 85% of all test samples had color errors of less than 1.0 ΔE2000 units, much more accurate than could be achieved by regression.