A novel method of face detection based on fusing YCbCr and HIS color space

Abstract

A face detection method for the color image with complex background is presented, which is a mixed skin-color segmentation model in both YCbCr and HIS color space constructed; The mixed skin-color segmentation model can be used to segment of possible skin regions in the original image. After skin-color segmentation, the geometrical shape information of face and the maximum of the complexity of eyes in faces can be used to detect face's position accurately. Experimental results show that the proposed approach can effectively prevent the side effects on the image caused by light condition, reduce the undetected rate, increase the speed of the face detection. Compared with other methods, this method has a very good result.Index Terms -face detection; YCbCr; HIS; complexity of images; feature 1ocation Ⅰ. INTRODUCTION Face detection is one of the most active research areas with wide range of interesting applications such as security, face recognition, human-computer interaction, surveillance etc. At present, the commonly used method of face detection can be summed up as knowledge-based, based on structural features, based on template matching and statistical model-based approach [1]. To be color images, using information of skin-color to detect human skin regions is a quick and effective way while narrowing the follow-up search space. As we know, the distance from the edge of the hair to the eyebrow, from the eyebrow to the tip of the nose, from the tip of the nose to chin, is equal; the most wide of the front of the face is the width of the five eyes; So it is capable of detecting face regions quickly and accurately based on the position of eyes. However, by the effects of illumination, skin color region probably is made as non-skin color region, that is unfavorable for skin-color detection and face location. In order to reduce the impact of brightness, normally use nonlinear YCbCr elliptic cluster skin-color segmentation model [2];the HS skin-color segmentation based on HIS color model etc. These methods reduce illumination effect of images to a certain extent. But some limitations still exist, such as that the segmentation effect of the images is bad which have great illumination or big reflection on faces. In this paper, we present a new method to get skin-color segmentation based on mixing nonlinear YCbCr elliptic cluster skin-color segmentation model and HIS skin-color segmentation model. After skin-color segmentation, the geometrical shape information of face and the maximum of the complexity of eyes in faces can be used to detect face position accurately. Ⅱ. SKIN-COLOR EXTRACTION A. Skin-color Model 1) YCbCr Skin-color Segmentation ModelTo be color images, the information of skin-color is very important characteristics for human face. Research shows that: even though of different races, different ages and different gender, the difference in color chrominance is far less than the difference in the brightness. Skin distribution shows clustering distribution in the skin-color space without luminance influence. Th...

Identifying Sport Types and Postures with Complex Background by Fusion of Local Descriptors

“…Previously, the intensity of color has been used as features in recognition problems. Color highlighting technique based on YCbCr color model 4,7,20,28,30,33,36 was used to emphasize blue and red colors for human skin detection and sport uniform color detection. Moreover, detecting foreground object by value channels of HSV color model was applied.…”

Color model selection for underwater object recognition

“…The choice of a color model is of great importance for underwater object recognition. Many task about color recognition have been done before, see e.g [7,8,9] etc. Based on dichromatic reflection color model [10], Gevers and Smeulders [11] provided several color models invariant to significant changes in viewpoint, object geometry and illumination.…”