Automatic measurement system of growth factor using bioassay of Hydra

1.九头蛇图像的二值化处理：九头蛇触角中球状队形反应的模式识别需要基于由CCD摄像机拍摄的原始九头蛇图像的二值化图像。对于九头蛇图像的二值化处理，我们提出了两种方法。其中之一使用Sobel过滤器来提取九头蛇的轮廓线。将具有最长轮廓线的级别确定为对九头蛇进行二值化处理的阈值。另一种方法是使用模拟人类视觉系统的神经网络。由于后一种方法的计算时间比前一种方法短，所以在九头蛇光照条件较好的情况下，采用了后一种方法。然而，在光照条件不是很好的情况下，我们使用了前者，因为它在九头蛇图像的二值化处理中具有很好的特性。触角球状反应的特征提取：我们使用了面积、轮廓线长度、圆度和九头蛇的分形维等特征来进行…更多的人认识到触手的球队形反应。用面积、轮廓线长度、圆度和分维对球型响应的识别率分别为64.5%、87.7%、93.0%和86.3%。将这些特征与触角长度特征相结合，识别率提高到93.2%。自动调换药室：我们使用井扫描仪在显微镜下调换药室。扫描仪由计算机通过RS232C串口控制。此外，从覆盖整个房间的680x480像素的图像中提取了320x240像素的图像。320x240像素的图像中心始终包含一条九头蛇。因此，即使九头蛇在室内的某个地方移动，也可以通过320x240像素的图像跟踪九头蛇。振动刺激增强触须的球队形反应：振动刺激增强触须的球队形反应。振动刺激的频率在27赫兹到31赫兹之间。较少

1. Binary processing of Hydra images : Pattern recognition of ball formation response in Hydra tentacles requires binary images based on original Hydra images taken by a CCD camera. For binary processing of Hydra image, we proposed two methods. One of them uses a Sobel filter to extract a contour line of Hydra. A threshold for binary processing of Hydra is determined as the level which has the longest contour line. The other method uses a neural network which mimics a human visual system. Since the calculation time was shorter in the latter one than in the former one, we used the latter one when Hydra has a good condition in illumination. In the case in which the condition of illumination was not so good, however, we used the former because of excellent characteristics in binary processing of Hydra images.2. Feature extraction for ball formation response of tentacles : We used features such as an area, a length of contour line, a degree of roundness and a fractal dimension of Hydra to … More recognize the ball formation response of tentacles. The recognition ratio of the ball formation response using the area, the length of contour line, the degree of roundness and the fractal dimension was 64.5%, 87.7%, 93.0% and 86.3%, respectively. A combination of these features and an other feature, a length of the tentacles, increased the recognition ratio to 93.2%.3. Automatic exchange of chambers : We used a well scanner to exchange a chamber under a microscope. The scanner was controlled by a computer through RS232C. In addition, an image with 320x240 pixel was extracted from an image with 680x480 pixel which covers a whole chamber. The image with 320x240 pixel always included a Hydra in the center. Thus, it was possible to track the hydra through the image with 320x240 pixel even though the Hydra moved somewhere in the chamber.4. Augmentation of ball formation response of tentacles by vibration stimuli : The ball formation response of tentacles was augmented by vibration stimuli. The frequency of the vibration stimuli was in the range of 27 Hz to 31 Hz. Less