Histogram equalization

About: Histogram equalization is a research topic. Over the lifetime, 5755 publications have been published within this topic receiving 89313 citations.

Histogram-Based Locality-Preserving Contrast Enhancement

Abstract: Histogram equalization (HE), a simple contrast enhancement (CE) method, tends to show excessive enhancement and gives unnatural artifacts on images with high peaks in their histograms. Histogram-based CE methods have been proposed in order to overcome the drawback of HE, however, they do not always give good enhancement results. In this letter, a histogram-based locality-preserving CE method is proposed. The proposed method is formulated as an optimization problem to preserve localities of the histogram for performing image CE. The locality-preserving property makes the histogram shape of the enhanced image to be similar to that of the original image. Experimental results show that the proposed histogram-based method gives output images with graceful CE on which existing methods give unnatural results.

Automatic color correction for sequences of images

Abstract: Automatic color correction is applied to a scene or clip, including a sequence of images, in a motion picture by selecting a representative image of the scene, analyzing the image and adjusting parameters of a color correction operation that is performed on the sequence of images included in the scene. This operation can be repeated automatically for all scenes or for selected scenes in the motion picture. The parameters may be adjusted to automatically color balance the image while maintaining substantially constant contrast. Analysis of the representative image may include identifying an offset of a peak in a two-dimensional histogram of the colors in the representative image from a white point. Parameters of a color correction operation are adjusted according to this offset. Separate histograms and offsets may be determined for shadows, midtones and highlight regions of the representative image. Analysis of the representative image may include determining a one-dimensional histogram of the luminance information in the representative image. The darkest level and the brightest level in the image are used to balance the image. In particular, the histograms for color channels in the image, such as red, green and blue, are adjusted to match the darkest level and brightest level identified by the luminance histogram.

Segmentation of Color Image Using EM algorithm in HSV Color Space

Abstract: This paper presents a new unsupervised method based on the Expectation-Maximization (EM) algorithm that we apply for color image segmentation. The method firstly Convert Image from RGB Color Space to HSV Color Space; Secondly we make use of a model of mixture K Gaussians, the Expectation Maximization (EM) formula is used to estimate the parameters of the Gaussian Mixture Model (GMM), which the desired number of partitions and fits the image histogram using a mixture of Gaussian distributions and provides a classified image; Thirdly, those pixels that have similar features will be regarded a group; Finally, for each group we segment pixels again according to their positions and we can get segmentation regions of the image. Experiment shows this method has better segmentation performance. The results of our methods are separately segmented and their combination allows the color image to be eventually partitioned.

Baseline CNN structure analysis for facial expression recognition

Abstract: We present a baseline convolutional neural network (CNN) structure and image preprocessing methodology to improve facial expression recognition algorithm using CNN. To analyze the most efficient network structure, we investigated four network structures that are known to show good performance in facial expression recognition. Moreover, we also investigated the effect of input image preprocessing methods. Five types of data input (raw, histogram equalization, isotropic smoothing, diffusion-based normalization, difference of Gaussian) were tested, and the accuracy was compared. We trained 20 different CNN models (4 networks x 5 data input types) and verified the performance of each network with test images from five different databases. The experiment result showed that a three-layer structure consisting of a simple convolutional and a max pooling layer with histogram equalization image input was the most efficient. We describe the detailed training procedure and analyze the result of the test accuracy based on considerable observation.

Family histogram based techniques for detection of commercials and other video content

Abstract: Techniques are disclosed for detecting commercials or other particular types of video content in a video signal. In an illustrative embodiment, color histograms are extracted from frames of the video signal. For each of at least a subset of the extracted color histograms, the extracted color histogram is compared to a family histogram. If the extracted color histogram falls within a specified range of the family histogram, the family histogram is updated to include the extracted color histogram as a new member. If the extracted color histogram does not fall within the specified range of the family histogram, the family histogram is considered complete and the extracted color histogram is utilized to generate a new family histogram for use in processing subsequent extracted color histograms. The resulting family histograms are utilized to detect commercials or other particular type of video content in the video signal.